Horticulture is one of the main pathways of deliberate introduction of non-native plants, some of which might become invasive. Of the 914 commercial ornamental outdoor plant species sold in Spain, 700 (77%) are non-native (archaeophytes excluded) marketed species. We classified these into six different lists based on their invasion status in Spain and elsewhere, their climatic suitability in Spain and their potential environmental and socioeconomic impacts. We found sufficient information for 270 species. We provide a Priority List of eight regulated invasive species that were still available on the market. We also established an Attention List with 68 non-regulated invasive and potentially invasive species that might cause various impacts. To prioritise the species within the Attention List, we further assessed the risk of invasion of these species by using an adaptation of the Australian WRA protocol and the level of societal interest estimated from values of the Google Trends tool. We also propose a Green List of seven species with probably no potential to become invasive, a Watch List with 27 potentially invasive species with few potential impacts and an Uncertainty List with 161 species of known status but with insufficient information to include them in any of the previous lists. We did not find sufficient information for 430 (61%) of the marketed non-native plant species, which were compiled into a Data Deficient List. Our findings of prohibited species for sale highlight the need for stronger enforcement of the regulations on invasive plant species in Spain. In addition, our results highlight the need for additional information on potential impacts and climate suitability of horticultural plants being sold in Spain, as insufficient information could be found to assess the invasion risk for most species.

e-commerce, Google trends, horizon-scanning, impact assessment, non-native plants, nursery, priority list, risk assessment

The introduction of invasive non-native species by humans may be accidental – for example seed crops as contaminants amongst cargo – or deliberate (Mack et al. 2000), such as the introduction of species used in forestry, aquaculture and horticulture. In particular, the sale of ornamental plants, including sale by nurseries, is the main deliberate pathway for plant invasions (Van Kleunen et al. 2018) which includes some of the most harmful invasive plant species in the wild (Hulme 2007).

There is a close relationship between domestic market-based propagule pressure and invasion success (Dehnen-Schmutz et al. 2007a; Johnston et al. 2009; Blackburn et al. 2013; García-Díaz et al. 2015). Horticulture activities reduce biotic and abiotic stresses on plants, bring species of different geographic origin together and increase the likelihood that plants escape into the wild (Niinemets and Peñuelas 2008). Moreover, introduction biases, or preferences for non-native species that perform better than natives (Chrobock et al. 2011), include plants that have increased germination rates, faster and larger growth and higher fecundity than native plants (Chrobock et al. 2011; Parker et al. 2013; Maurel et al. 2016). Therefore, the commercial use of non-native ornamental plant species is not only important as the main pathway of introduction (Hulme 2007), but it also favours the invasion potential of these plants and their impacts. In fact, in less than 20 years on the market, some non-native species can become invasive (Pemberton and Liu 2009).

The most effective way to manage the impacts of non-native species is through prevention (Convention on Biological Diversity 2010). In the last two decades, there has been great progress in developing risk assessment protocols as an essential management component to identify potentially invasive species (Pheloung et al. 1999; Leung et al. 2012; Roy et al. 2014, 2015). Most of these risk assessments are used to rank non-native species according to their probability of becoming established and causing harm. Horizon-scanning of invasive species is a particular type of rapid screening risk analysis based on the systematic examination of future potential threats, leading to the prioritisation of non-native species for further investigation (Roy et al. 2014). Horizon-scanning has been applied when prioritisation identifies a small fraction of species selected from a large list of scanned species, for which a thorough risk analysis is not feasible in a short period of time (Andreu and Vilà 2010; Roy et al. 2015; Roy et al. 2019). This is the case, for example, for testing the invasion risk of all ornamental plants commercialised within a country because the number of non-native plant taxa sold is very high.

A parsimonious way to perform a horizon-scanning analysis for ornamental plants is to use four of the most widely used criteria to identify potentially invasive species: climate matching, being invasive elsewhere, their potential impacts on the environment and their impacts on socioeconomic activities (Weber and Gut 2004; Otfinowski et al. 2007, Gassó et al. 2010; Blackburn et al. 2014; Roy et al. 2014).

As the number of non-native species being used as ornamentals is very high, but resources are limited to manage them all in the same way, it is necessary to create prioritisation lists of plant species identifying those that are (1) regulated invasive but still commercialised, (2) potentially invasive with the risk of causing numerous impacts, (3) potentially invasive with few potential impacts, (4) probably safe because there is no potential to establish in the wild and (5) those for which there is insufficient information to classify them by their risk of invasion and impact (Dehnen-Schmutz 2011). This coarse screening is the basis for prioritising the potentially most invasive species and to later perform a more detailed risk assessment (Pheloung et al. 1999; Weber and Gut 2004; Andreu and Vilà 2010; Gassó et al. 2010), as well as for proposing a list of the least harmful species (Gederas et al. 2012).

Since many regulations expressly prohibit the commercialisation of listed species, it is expected that none of them is sold. However, regulation is not always effective, especially for the online plant trade (Humair et al. 2015). In fact, many nurseries continue to stock and supply invasive regulated species (Wirth et al. 2004; Cronin et al. 2017). In addition, nowadays most of the nurseries offer internet purchasing. This model of commerce is currently one of the most important sources of gardening plants (Humair et al. 2015). The online plant trade significantly increases transportation distance and propagule pressure of non-native species (Walters et al. 2006; Lenda et al. 2014; Humair et al. 2015). As an innovative approach, the Google Trends tool can be used to forecast consumption and commerce, which is a valuable source of information (Vosen and Schmidt 2011). Therefore, the level of interest in each ornamental species measured by Google Trends can provide information on the interest for a particular species by society in general. Google Trends provides information on how frequently a keyword or group of keywords has been searched for on the Internet. Resulting data are not necessarily composed of only people interested in buying the plant. In fact, it is possible that some of these searches are performed because they are looking for ways to control or manage already invasive species. In our study, we used Google Trends data to assess "popularity" or "interest", defined in a broad sense, as the data do not allow distinguishing whether plant name searches were motivated by Internet users’ positive or negative views of a plant.

To our knowledge, this tool has not yet been used for the prevention of biological invasions.

In this paper, we perform a horizon-scanning analysis of the 914 commercial ornamental outdoor plant species in Spain from a total of 1063 taxa to facilitate policy implementation. The main aim is to generate six species lists based on their regulation and invasive status in Spain and elsewhere, climate matching between their native region and Spain, the magnitude of the environmental and socioeconomic impacts they might cause and their societal interest (Fig. 1):

Figure 1.

Flow diagram of horizon-scanning of commercial ornamental non-native plant species in Spain and their classification into respective lists. Colour codes correspond to those in Table 1.

• A Priority List that includes regulated (by Spain or the EU) invasive non-native species that were still commercially available in the Spanish peninsular territory (Spain, hereafter).

• An Attention List that includes climatically suitable non-regulated invasive in Spain and potentially invasive species (i.e. invasive elsewhere) with many potential impacts.

• A Watch List that includes climatically suitable non-regulated invasive in Spain and potentially invasive species (i.e. invasive elsewhere) with few potential impacts.

• A Green List that includes species with no climatic suitability and probably no potential to be invasive in Spain.

• An Uncertainty List that includes non-invasive species with probably no potential to be invasive that do not meet the requirements to be included in the Green List. It also includes species with known invasion status but with insufficient information available on impacts and non-native species with known invasion status but with insufficient information on climatic suitability or invasiveness elsewhere.

• A Data Deficient List with all the non-native species with no information about their invasion status and not enough data to classify them in any other list.

To rank the species of the Attention List, we conducted an in-depth analysis based on their risk of invasion and societal interest. The risk of invasion was scored according to the Australian weed risk assessment performed by Pheloung et al. (WRA 1999). This WRA protocol has been tested successfully for its consistent accuracy in different geographic regions (Gordon et al. 2008) including Spain (Gassó et al. 2010). Societal interest in non-native species was measured using Google Trends. We wanted to answer the following questions: Does the risk of invasion and interest to society grow across the different invasion status groups of species currently in Spain (i.e. not in the wild, casual, naturalised and invasive)? Does the number of potential environmental and socioeconomic impacts rise with increasing invasion status. Moreover, for each non-native species, we calculated a Priority Index based on the impacts, WRA score and interest of the species. We tested if the Priority Index increased with increasing invasion status of the species.

We compiled a database including the vast majority of ornamental outdoor plants with commercial use in gardening in the Spanish peninsular territory excluding the Canary and Balearic Islands (Spain, hereafter). We also included indoor plants that can survive and/or reproduce outdoors. However, we excluded strictly indoor plants because their ecological requirements might prevent survival outdoors. We included fruit trees as they are of ornamental use in public and private gardens and green areas, but we excluded vegetables used in horticulture. The list of taxa was compiled through the systematic consultation of catalogues from the 21 main Spanish nurseries (Appendix 1) between December 2015 and October 2016, which provide plants for sale across the country. The number of new taxa added to the database (Appendix 1) did not increase at all after the 15^th nursery catalogue was consulted. The total number of taxa compiled was 1036, of which 914 were actual species. Hybrids or genus level taxa were not included in the analysis and infraspecific taxa were pooled into species.

According to their origin, we first discriminated between native species and non-native species in Spain. We then identified archaeophytes (i.e. species introduced before 1500 A.D.), following Pyšek et al. (2004). Archaeophytes were not included in the analysis because they are poorly recorded and, for many species, their non-native status is under discussion.

Species were further classified according to their invasion status in Spain, following the definitions recommended by Richardson et al. (2000): not in the wild, casual, established or naturalised (hereafter "naturalised"), invasive non-regulated in Spain or Europe (hereafter "invasive") or invasive regulated in Spain or Europe (hereafter "regulated"). Invasion status in Spain, as not in the wild, casual, naturalised or invasive, was based on the information provided in the Spanish Atlas of Invasive Non-native Plants (Sanz Elorza et al. 2004). The regulation status of the species, that in Spain involves the ban of possession, transport and commerce of living beings and propagules, was based on the Spanish Catalogue of Non-native Invasive Species (BOE 2013) and the List of Invasive Alien Species of Union Concern (European Commission 2016, 2017).

Once the non-native species were classified into these five invasion status groups (i.e. regulated invasive, invasive, naturalised, casual, not in the wild), we proceeded to perform the horizon-scanning to classify the species into the respective lists based on the flow diagram illustrated in Fig. 1 as follows:

All Regulated species were directly included in the Priority List, whereas invasive species were considered for impact assessment (see method below).

Naturalised species were identified as invasive elsewhere, based on the CABI Datasheets (2018) and the Global Invasive Species Database (Invasive Species Specialist Group 2015). Naturalised species that are invasive elsewhere were regarded as potentially invasive species and were considered for impact assessment. Naturalised species not invasive elsewhere were included in the Uncertainty List.

Casual and not in the wild species were screened for climatic suitability in Spain (see method below). For species climatically suitable somewhere in the country, with a medium to high level of confidence in the likelihood, we checked whether they were invasive elsewhere (Roy et al. 2014). Species that were not climatically suitable and not registered as invasive elsewhere, were included in the Green List (Dehnen-Schmutz 2011), commonly known as the "white list" by decision-makers. In contrast, species that were climatically suitable and were invasive elsewhere were regarded as potentially invasive species and thus considered for impact assessment. On the other hand, species that were climatically suitable but not invasive elsewhere and species that were invasive elsewhere but not climatically suitable were included in the Uncertainty List.

Finally, all the species naturalised or casual in Spain, for which we could not find sufficient data about being invasive elsewhere and those for which we have a low level of confidence in the likelihood of climatic suitability or no data at all, were included in the Uncertainty List. This list also includes species not in the wild that are invasive elsewhere rather than Spain, species that are not climatic suitable or species in which the level of confidence in the likelihood of climatic suitability is very low.

In contrast, the Data deficient list includes species not in the wild, with no data on status elsewhere and on climate suitability.

The data underpinning the analysis, reported in this paper, are deposited in the Zenodo repository at https://doi.org/10.5281/zenodo.3367257 (Bayon and Vilà 2019).

Of the 914 taxa identified to species, 199 were native to Spain and 15 were archaeophytes. Of the 700 remaining non-native species, we did not find sufficient information on invasion status, climatic suitability or invasiveness elsewhere for 430 species (Data deficient list; Appendix 4) in the consulted databases (Invasive Species Specialist Group 2015; CABI 2018). For the remaining 270 non-native species, 71 taxa were not in the wild, 99 were casual, 70 naturalised and 30 invasive in Spain. Of the invasive species in Spain, eight are regulated by the Spanish Catalogue of Non-native Invasive Species (BOE 2013) and one of them, Pennisetum setaceum, is also regulated by the List of Invasive Alien Species of Union Concern (2016, 2017).

Nineteen species not in the wild, 30 casual and 24 naturalised species are climatically suitable and invasive elsewhere and thus considered potential invaders. These species, in addition to the 22 already invasive non-regulated species, were assessed for impact (Table 1).

Impact Assessment

We assessed the potential impact of the above-mentioned 19 not in the wild, 30 casual, 24 naturalised and 22 invasive species that are climatically suitable and invasive elsewhere. The global median value for environmental impacts was three and the median for socioeconomic impacts was one. Therefore, species with impacts at or above these values were included in the Attention List. This included eleven not in the wild (58%), 22 casual (73%), 20 naturalised (83%) and 15 invasive species (68%). We did not find significant differences in the number of environmental or socioeconomic impacts across any pair of species status groups (Fig. 2).

Figure 2.

Environmental (a) and socioeconomic (b) impacts of invasive and potentially invasive ornamental plant species. P-values for Tukey’s range tests for environmental impacts: not in the wild – casual: p = 0.833; not in the wild – naturalised: p = 0.498; not in the wild – invasive: p = 0.926; casual – naturalised: p = 0.904; casual – invasive: p = 0.997; naturalised – invasive: p = 0.845. P-values for Tukey’s range tests for socioeconomic impacts: not in the wild – casual: p = 0.790; not in the wild – naturalised: p = 0.526; not in the wild – invasive: p = 0.916; casual – naturalised: p = 0.947; casual – invasive: p = 0.994; naturalised – invasive: p = 0.875. Dashed red line represents the global median of impacts and the threshold for species in the Attention List (at and above the line, Table 3) and Watch List (below the line, Appendix 2)

Within each invasion status, the species with the highest number of environmental impacts included the invasive Robinia pseudoacacia (7), Agave sisalana (6) and Elaeagnus angustifolia (6); the naturalised Ficus pumila (7), Ficus rubiginosa (6), Lupinus polyphyllus (6) and Sansevieria trifasciata (6); the casual Wisteria sinensis (8) and Grevillea robusta (7); and the not in the wild Nymphaea odorata (7).

Species with the highest number of socioeconomic impacts were the invasive Robinia pseudoacacia (3), Acacia longifolia (3), Eucalyptus globulus (3) and Lantana camara (3); the naturalised Lupinus polyphyllus (3), Rhus typhina (3) and Tagetes minuta (3); the casual Miscanthus sinensis (3), Portulaca oleracea (3) and Sesbania punicea (3); and the not in the wild Allamanda cathartica (3).

Societal interest analysis of Attention List species

In the Google Trends systematic examination of the 68 Attention List species, maximum trend values were observed for Robinia pseudoacacia – March 2004 – and therefore we used this record as our control species.

Within the Attention list, the most noteworthy species (higher STV) included: the invasive Robinia pseudoacacia (100) and Lantana camara (67); the naturalised Phoenix canariensis (71) and Melia azedarach (71); the casual Cestrum nocturnum (80) and Zantedeschia aethiopica (70); and, far from the previous groups, the not in the wild Leptospermum scoparium (43). Complete results of the STV analysis are shown in Table 3. There were no significant differences in STV between any pair of invasion status groups of species (Fig. 3). Similarly, there were no differences between the species in the Priority list and the Green List (p=0.967).

Figure 3.

Society interest in ornamental non-native plant species classified by their invasion status. STV: Standard Trend Value. P-values for Tukey’s range tests: not in the wild – casual: p = 0.373; not in the wild – naturalised: p = 0.783; not in the wild – invasive: p = 0.436; casual – naturalised: p = 0.794; casual – invasive: p = 1; naturalised – invasive: p = 0.860.

Invasion risk assessment of Attention List species

In the WRA, all species in the Attention List were rejected, except for 11 that required further evaluation and only one, Annona cherimola, that was accepted (Table 3).

Within each invasion status, the highest WRA scores were for the invasive Lantana camara (25), Acacia longifolia (23), Acacia saligna (22), Elaeagnus angustifolia (21), Eucalyptus globulus (21) and Leucana leucocephala (21); the naturalised Phragmites australis (27), Lupinus polyphyllus (27), Pennistum villosum (25), Canna indica (24) and Imperata cylindrica (24); the casual Kalanchoe daigremontiana (22), Pawlownia tomentosa (19) and Eugenia uniflora (18); and the not in the wild Cotoneaster horizontalis (26), Cornus sericea (22) and Physalis angulata (22). For every invasion status, the species requiring further evaluation accounted for less than 25%. There were no significant differences in WRA scores between any pair of invasion status groups of species (Fig. 4).

Figure 4.

Weed risk assessment (WRA) score in ornamental non-native plant species classified by their invasion status in Spain. P-values for Tukey’s range tests: not in the wild – casual: p = 0.392; not in the wild – naturalised: p = 0.983; not in the wild – invasive: p = 0.951; casual – naturalised: p = 0.101; casual – invasive: p = 0.086; naturalised – invasive: p = 0.997.

Prioritisation of Attention List species

The median value of Priority Indices was 35. Species with a Priority Index ≥ 35 are highlighted in Table 3. Within each invasion status, the highest Priority Indices in invasive species were found for Robinia pseudoacacia (73), Lantana camara (68) and Eucalyptus globulus (55); in naturalised species Lupinus polyphyllus (56) and Canna indica (54); in casual species Portulaca oleracea (54), Cestrum nocturnum (49) and Wisteria sinensis (49); and in not in the wild species Cinnamimum camphora (45) and Cotoneaster horizontalis (45). There were no significant differences between any pair of invasion status groups of species (Fig. 5).

Figure 5.

Priority Index in ornamental non-native plant species classified by their invasion status in Spain. P-values for Tukey’s range tests: not in the wild – casual: p = 0.981; not in the wild – naturalised: p = 0.860; not in the wild – invasive: p = 0.633; casual – naturalised: p = 0.958; casual – invasive: p = 0.748; naturalised – invasive: p = 0.953. Dashed red line represents the global median of Priority Index (=38). Species at or above this line should be considered in prioritisation, as shown in table 3.

Nurseries and the commercial introduction of non-native plant species are the main deliberate pathways for plant invasions (Van Kleunen et al. 2018). Some of the most harmful invasive plant species in the wild are non-native species introduced for commercial purposes (Hulme 2007). In Spain, non-native plants represent the vast majority of species sold by nurseries (77%) and 30 of these species have been reported as invasive in the peninsular territory of Spain. The regulation of non-native invasive plant species is necessary. However, nurseries often do not fully comply with commercial restrictions (Wirth et al. 2004; Cronin et al. 2017; Touza et al. 2014). Besides the Spanish (BOE 2013) and European (European Commission 2016, 2017) regulations on non-native invasive species and similar to what happens in other countries, there are eight regulated species that, although being regulated, were still commercially available in the country at the time of this study (Appendix 1), as indicated in the Priority List. One of them, Pennisetum setaceum, was also been included under European regulation, although this inclusion is more recent than the study of the nurseries for this study. Therefore, there is a pressing need to enforce the current legislation, as well as to raise public awareness to prohibit the trade of these species.

Furthermore, because the deliberate transport, commerce and planting of non-native plant species can be controlled, all the invasive and potentially invasive species compiled into the Attention List could be considered for regulation, following the advice of the European Parliament and Council (2014). While some of these species have been introduced to provide an immediate economic benefit, such as Eucalyptus globulus and other species of the same genus (Touza et al. 2014), their impacts on the environment, as well as on some human activities besides the forestry sector, suggest that their regulation should be considered. In addition, the costs derived from the control of invasive species can be quite significant (Pimentel et al. 2005) and are not compensated for by their economic benefits.

The levels of potential impacts of species in the Attention List are independent of their invasion status in Spain. That is, current non-invasive species have the potential to cause as many impacts as invasive species. This result supports previous empirical studies indicating that invasiveness does not always translate to impacts (Ricciardi and Cohen 2007). Despite the fact that some of these species were introduced a long time ago and are already invasive, such as Ailanthus altissima introduced into Spain in 1818 or Acacia dealbata in 1824 (Sanz Elorza et al. 2004), others have possibly been introduced recently as ornamental plants and thus have not had enough time for establishment and dispersal (Pemberton and Liu 2009). As a precautionary measure to avoid their impacts, there is a need to prevent the introduction of any non-native species listed in the Attention List and to conduct a complete risk assessment to study the possibility of their being regulated.

As the threshold number of environmental and socioeconomic impacts required for a species to be included or not in the Attention List is based on median values, it may exclude some species that have high risks of invasion despite a low number of impact types or species with still unknown impacts. In fact, the Watch List contains species that are well known to be invasive elsewhere, such as Acer negundo or Eriobotrya japonica, which cause few, but important, impacts. More detailed research on the potential type of impacts of the species in the Watch List, as well as conducting a WRA for these species, would allow for more adequate prioritisation of these species that are of major concern after those in the Priority and Attention Lists.

We are confident that the potential invasion of Attention List species is robust given the positive results of the WRA, in which only one of the 68 species listed was classified as accepted (i.e. low invasion risk). The proportion of species rejected by the WRA was very high and similar in all status groups of species, with a likely low incorporation of false positives (Andreu and Vilà 2010). Furthermore, our WRA analysis warns that species which are not in the wild, still have the potential to become invasive, reaching scores as high as those of already naturalised or invasive plants.

Likewise, with our analysis on the STV, we cannot infer causality between societal interest and increased commerce. In fact, the STV is not a good predictor of invasion status; the absence of differences between STV of the species in the Priority list and the Green list shows that the STV is not related with invasion. Nevertheless, greater interest, as reflected in Google statistics, can still be an indicator of increased consumption and trade (Vosen and Schmidt 2011) and greater interest might promote greater propagule pressure and greater potential to become established (Dehnen-Schmutz et al. 2007a; Johnston et al. 2009; Blackburn et al. 2013; García-Díaz et al. 2015). Thus, we think that the STV continues to have value, not as a predictor of invasion, but as a factor to be considered during prioritisation of the species within each invasion group.

Our approach is similar to previous Horizon-scanning analyses for non-native species (Roy et al. 2014, Roy et al. 2019) with the differences being that, in our study case, all the species are already in the region of analysis and the scoring is based on more parameters than just their potential to establish and cause impacts. We provide a Priority Index that includes the risk of invasion, the level of potential impacts and the popularity of the species. Therefore, it offers an integrative score that may be of interest to environmental administrations and management services as a valuable tool to support decision-making. The homogeneity in the Priority Index across the invasion status groups of species confirms that the potential for invasion and the impact risks are independent of the actual invasion status of the species (Roberts et al. 2011; Gassó et al. 2010). Therefore, this index may even be useful for identifying current non-invasive species that, even if not in the wild, are potentially invasive. Indeed, our species Prioritisation List has already been used to identify species with the potential to be invasive in Gibraltar, an overseas territory of the United Kingdom which buys all its ornamental plants from neighbouring Spain according to the UKOTs Horizon Scanning and Biosecurity Workshop that took place on 21–24 January 2019 (K. Bensusan, pers. Com).

We also generated an Uncertainty List composed of species that probably do not represent an immediate invasion risk. For fifty-six percent of the species listed, we lack sufficient confidence in the likelihood of climatic suitability (or we have no information about it). Even if the species in the Uncertainty List do not become established in Spain due to their climatic requirements, or they are not invasive elsewhere, we need to be aware that these two criteria can change over time. For example, a particular climate change scenario could cause climatically unsuitable species to become suitable in the future (Mainka and Howard 2010).

Unfortunately, we could not find information on the status, invasive potential and climatic suitability of 61% of the non-native species sold in nurseries. There is a worrying possibility that the Data Deficient List includes some potentially invasive species that are not considered in the two major databases consulted (CABI Datasheets 2018; Invasive Species Specialist Group 2015). Additional research, as well as consultation of the primary literature, is needed to allow reclassifying species from the Data Deficient List, a task that would require the expertise of a larger team of scientists (Roy et al. 2019; González-Moreno et al. 2019). The immediate task would be to use the new available GLONAF database to identify invasive species elsewhere (van Kleunen et al. 2019).

Finally, we provide a Green List of non-native species with very low invasion potential. Promoting preferences for non-invasive species in horticulture can be a valuable endeavour in order to make regulations easier to comply with (Gagliardi and Brand 2007). Involving the horticultural industry in the dissemination of plant invasion risks and in the development of regulations has been shown to be effective (Humair et al. 2014). This Green List can be a starting point for the establishment of voluntary codes of conduct amongst nursery owners (Reichard 2004; Gagliardi and Brand 2007; Robinson et al. 2017). However, it is prudent not to forget that propagule pressure is an important factor determining invasion (Lockwood et al. 2005; 2009; Johnston et al. 2009).

If species in the Green List are planted frequently, in large quantities and in many locations, this scenario can be changed. For this reason, the Green List presented here is short and tentative; further and more in-depth research is needed on the Uncertainty List so as to possibly enlarge this Green List. Planting native species will always be the preferable alternative.

The present research is preliminary in nature and the authors are aware of the clear limitations of the conclusions. However, we consider it can be a very useful and complete tool to establish priorities in long lists of species for which not much information is available and it represents a good starting point for more thorough and detailed risk analyses that allow the improvement and implementation of new and more efficient forms of regulation of invasive species.

Based on our prioritization list analysis, we provide the following recommendations: 1) there is a need to reinforce the current legislation and implement systems that guarantee its compliance regarding the species of the Priority List; 2) invasive species in the Attention List should be considered for regulation; 3) established, casual and not in the wild species in the Attention List, especially those with higher Priority Index values, should be included in a monitoring programme to prevent future invasions; 4) species in the Watch List should be included in an early warning programme if they are not yet in the wild and monitored if they are already established in a few localities; 5) species in the Uncertainty List require further evaluation in order to be reclassified into Attention, Watch or Green Lists; finally 6) species in Data Deficient List require further information on their status, invasive potential or climatic suitability in order to be reclassified into Attention, Watch or Green lists.

This research also reveals the limited responsibility and awareness by some commercial nurseries regarding the sale of invasive species. A better monitoring and tracking system for the species for sale and more rigorous inspections in nurseries are very necessary (Touza et al. 2014). Citizen awareness through the dissemination of knowledge about invasive non-native species, as well as citizen science projects working with gardeners, can be useful tools to reduce their demand and consumption as suggested in other studies (Reichard 2004; Gagliardi and Brand 2007; Robinson et al. 2017; Dehnen-Schmutz and Conroy 2018). Having a national registry of ornamental plant species available for sale and the requirement of a risk analysis for the introduction of new species in it, is also a recommendation to be considered.

This study was funded by the project IMPLANTIN (CGL2015-65346-R). Á. Bayón is hired by the Biological Station of Doñana (EBD-CSIC) thanks to a Grant for Predoctoral Contracts for the Training of Doctors 2015 awarded by the Ministerio de Ciencia, Innovación y Universidades of Spain and co-financed by the European Social Fund (BES-2015-072929). We thank P. Castro, O. Godoy and F. Essl, S. Vanderhoeven, P. Hulme and the associated editor for comments on previous versions of the manuscript and J. Arroyo for tutoring the thesis project at the University of Seville.