Background & summary

The impact of biological invasions poses one of the greatest threats to nature1, with rapidly emerging shifts in species ranges and subsequent biodiversity loss impacting ecosystem services2, economies3 and human well-being4. Of special concern are alien plant invasions, because of the great number of species (~14,000) introduced outside their native range compared to other taxa5 with no sign of saturation6.

Alien plants can be classified based on their position along the introduction–naturalisation–invasion status continuum7,8. According to the unified invasion status framework proposed by Blackburn et al. (2011), species introduced in a new region are classified differently depending on the barriers they have overcome, regardless of their impact8. As such, casual aliens are those that have been transported beyond the limits of their native range. Only a fraction of casual aliens become naturalised to form self-sustaining populations. Finally, only a fraction of those naturalised become invasive, overcoming local dispersal barriers and spreading in the new region8.

High quality, easy-to-access, standardised and unified data sources are essential for assessing and monitoring the status of biological invasions, making future predictions and for prioritizing management actions. While there are many freely accessible online databases, their ultimate value depends on the feasibility of integrating information provided at different spatial scales, following different criteria, or using different nomenclature9.

Recently, there have been many initiatives to create and make available large-scale datasets of alien flora. Such is the case, for example, of the Global Naturalized Alien Flora (GloNAF)10 at the global scale, or the Inventory of alien invasive species in Europe (DAISIE)11.

Australia has the second highest number of endemic plant species in the world12 and is one of the world’s biodiversity hotspots and megadiverse countries, which are known to be the most vulnerable to biological invasions13. To prevent biological invasions, $13.6 billion are currently invested every year in environmental biosecurity in Australia14. Australia’s plant censuses, including information on whether a species is native or introduced, have been developed independently by state and territory government environment departments, which has resulted in inconsistencies at the national level. The only national sources about alien flora that exist in Australia are outdated or apply unclear criteria for species’ inclusion and status15,16,17 (http://wwf.org.au/publications/ListInvasivePlants/).

To overcome mismatches caused by jurisdictional boundaries and enable efficient management, standardised data including invasion status at the national scale are needed.

The most accessible and comprehensive plant census at the national scale is the Australian Plant Census (APC; https://biodiversity.org.au/nsl/services/export/index; accessed February 2022), a nationally accepted taxonomy for vascular plants (angiosperms, ferns and gymnosperms). It is one of the taxonomic resources of the Australian National Species List (auNSL; https://biodiversity.org.au/nsl/) and provides authoritative data for names and published taxon concepts for native and naturalised taxa in Australia. Despite being federally managed and endorsed by the Council of Heads of Australasian Herbaria (CHAH), the APC provides information on a state-by-state basis, but is not synthesised nationally. Recently, the Global Register of Introduced and Invasive Species (GRIIS) for Australia18 was published by the Invasive Species Specialist Group (ISSG)19 following the methodology proposed by Pagad et al.20. However, to date, Australia is still lacking a dataset that integrates all the existing data sources across the six Australian states and two main territories (for simplicity hereafter referred to as ‘states’) with those that have been produced at the national scale. In order to deliver standardised up-to-date information and create a unified dataset nationally, an automated system that integrates datasets and specifies plant species origin and introduction status is needed. This will provide a strong evidence-base for planning and informing actions for prevention and to mitigate risks21.

In this paper, we collate, combine and unify information from ten independent data sources for all alien plant species recorded in Australia into a unified and standardised national dataset, the Alien Flora of Australia (AFA). By integrating all existing data sources and embedding a technical validation step, this dataset is the first of its kind for Australia.

To construct the AFA, we first had to integrate and standardise terminology on plant invasions across Australia and propose prioritisation systems to tackle mismatches existing at different spatial scales22. Then, we had to develop (i) an R script able to combine the information contained within the APC and Australian state plant censuses, and (ii) updated standardised datasets at the jurisdictional level. Afterwards, we integrated all the information into (iii) the Alien Flora of Australia (AFA), a nationally unified dataset. With its associated script, the dataset presented here can be regularly updated with each new release of the state plant censuses. Hence, the Alien Flora of Australia and its associated script enable users to leverage both Australian state and national plant censuses and create customised data subsets suited to their specific needs regarding both area of interest and desired taxonomy. Through its real time coordination and reconciliation of open-access data, the approach used to create the AFA can help transform knowledge, understanding and the efficacy of invasive plant species ecology, management and policy in Australia.

Methods

We used R 4.1.1 (R Core Team; https://www.R-project.org/) to create an R script that combines the information recorded on the APC with all the state flora censuses and the GRIIS in order to detect mismatches between species nomenclature and introduction status. The script also allows addressing those mismatches into elucidating a unified introduction status for each species at the national level22. The script can be used with future releases of the APC or any of the state floras, to automate this process. There are seven main steps underpinning the creation of the AFA, which we outline here (Fig. 1).

Fig. 1
figure 1

The steps to create the Alien Flora of Australia (AFA) dataset, including obtaining the data sources, curation and standardisation, cross-reference information at the state level and unify introduction statuses at the state level, integrate introduction status information at the national level, cross-reference information at the national level and unify introduction statuses at the national level. APC is the acronym of the Australian Plant Census whereas GRIIS is the acronym of the Global Register of Introduced and Invasive Species. States and main territories have also been abbreviated (the Australian Capital Territory, ACT; New South Wales, NSW; the Northern Territory, NT; Queensland, QLD; South Australia, SA; Tasmania, TAS; Victoria, VIC; Western Australia, WA).

Full size image

Data sources (step 1)

The AFA harmonises data from ten independent repositories, which we have combined and integrated using the steps and framework detailed in Fig. 1. Firstly, we obtained the three major types of data sources to be used in the creation of the AFA either via direct download or obtained from authoritative sources. The data sources included: i) the Australian Plant Census (APC); ii) the state flora censuses (eight repositories in total); and iii) the Australian Global Registry of Introduced and Invasive Species (GRIIS) (Fig. 1; Tables 1–3). We only included information on vascular plants; other taxa were excluded at the first step of data curation.

Table 1 Sources for state vascular flora censuses, latest version available for each of them, or whether continuously updated, accessed date, and conversion procedure followed in order to standardise the introduction and naturalisation statuses for all states to make them comparable.
Full size table
Table 2 Sources for state vascular flora censuses, latest version available for each of them, or whether continuously updated, accessed date, and conversion procedure foll and naturalisation statuses for all states to make them comparable (Continuation of Table 1).
Full size table
Table 3 Sources for state vascular flora censuses, latest version available for each of them, or whether continuously updated, accessed date, and conversion procedure followed in order to standardise the introduction and naturalisation statuses for all states to make them comparable (Continuation of Tables 1 and 2).
Full size table

The first and third data sources, the APC and the GRIIS, are at the national scale. The Australian Plant Census (APC) is a list of the Australian flora, both native and introduced; it includes accepted names, synonyms and misapplications for these names, and specifies species distribution on a state-by-state basis for continental Australia and its external territories. Taxa known only from cultivation in Australia are excluded in the APC, and therefore also from the AFA database created here. The Australian GRIIS checklist is a species catalogue for introduced and invasive species recorded from Australia that includes taxonomic information based on the Global Biodiversity Information Facility (GBIF).

The second source of data consisted of the eight plant censuses developed independently by environmental government agencies and herbaria associated with each of the two main Australian territories (Australian Capital Territory (ACT; https://www.cpbr.gov.au/cpbr/ACT-census/vascular-gen-alpha.html) and the Northern Territory (NT; http://eflora.nt.gov.au/NTSpeciesList?heading=sNTSpecies)), and the six Australian states (New South Wales (NSW; https://plantnet.rbgsyd.nsw.gov.au), Queensland (QLD; https://www.data.qld.gov.au/dataset/census-of-the-queensland-flora-and-fungi-2022), South Australia (SA; https://data.environment.sa.gov.au/Content/Publications/vascular-plants-bdbsa-taxonomy.xlsx), Tasmania (TAS; https://flora.tmag.tas.gov.au/resources/census/), Victoria (VIC; https://vicflora.rbg.vic.gov.au) and Western Australia (WA; https://florabase.dbca.wa.gov.au/)) (see Tables 1–3 for details on information included within each state census).

The Alien Flora of Australia (AFA) is available for the latest release of the APC (2022) and the GRIIS v1.9 (2022). Similarly, the state-by-state comparative datasets created here are available for the latest releases of the state plant censuses, ranging from 2019 for the ACT, to the most updated ones for the rest of the states, corresponding to censuses as of 2022 (except for Victoria, where the latest release dated from 2023).

Curation and standardisation (step 2)

Herbaria from different states follow various versions of protocols for classification, and display the available data inconsistently, therefore limiting data federation. Thus, after obtaining all the data sources we had to standardise them (Fig. 1).

Standardised taxonomy

We followed the taxonomy and nomenclature adopted for the APC as the most recognised authority for the Australian vascular flora, which is endorsed by the Council of Heads of Australasian Herbaria (CHAH). In addition, our script allows matching the taxonomic nomenclature used in the APC, to other taxonomies extensively used internationally, such as the GBIF taxonomic backbone (https://doi.org/10.15468/39omei) and the standardised nomenclature provided by WorldFlora Online (WFO; http://www.worldfloraonline.org/) and provide all of them in the AFA. We cross-referenced the taxonomy and nomenclature with the one used in the GBIF taxonomic backbone and with WFO, using rgbif23 and WorldFlora24 R packages, respectively, without fuzzy matching.

Standardised terminology in invasion ecology

To address incompatibilities and differences regarding terminology on invasion ecology when cross-referencing information among the data sources, we based the AFA terminology on an adaptation from Blackburn’s framework8 described in detail by Martín-Forés et al.22. Such proposed terminology is the most directly comparable with the one employed on the APC, and it provides information about presence, origin and introduction status in a combined manner. Following the discussion addressed by Martín-Forés et al.22, native species in the AFA can have one of the following statuses: native, native colonising or native potentially colonising (when they were also naturalised or potentially naturalised, respectively within the same state or territory of origin). Alien species in the AFA can be considered: introduced (to refer to casual aliens or in cases where no information on naturalisation is available), naturalised (when forming self-sustaining populations) – it is stated whether, for the two former ones, the status was doubtfully or formerly when that was the case –, or harmful invasive (when recorded as invasive knowing to cause negative impacts). If the origin for a species is uncertain, it is recorded as ‘uncertain origin’, whereas taxa that are no longer present are recorded as ‘presumed extinct’ (see Tables 4, 5 for definitions).

Table 4 Glossary of terms used in the Alien Flora of Australia (AFA) adapted from Blackburn et al. (2011).
Full size table
Table 5 Glossary of terms used in the Alien Flora of Australia (AFA) adapted from Blackburn et al. (2011) (Continuation of Table 4).
Full size table

In addition, the AFA also includes the Darwin Core Standard (https://dwc.tdwg.org/) proposed by the Biodiversity Information Standards (TDWG) to refer to the species origin (establishment means; http://rs.tdwg.org/dwc/doc/em/2021-09-01), and the extent to which an introduced species survives, reproduces, and expands in its introduced range (degree of establishment; http://rs.tdwg.org/dwc/doc/doe/2021-09-01). By including both the Blackburn’s framework and the Darwin Core standard, we intend for the AFA to be as useful as possible for researchers, biosecurity departments, environmental managers, database managers and herbaria alike.

Curation and rearrangement of the Australian Plant Census

We downloaded the APC dataset from the Vascular Plants NSL Service APC20 and used the package stringr25 to work and manipulate text strings. The information on the column “Distribution” from the APC, included in a consecutive manner for each row all the information about a species presence and origin in all the Australian states and main and external territories in which it was present. We therefore split the information on distribution into individual columns corresponding to each of the Australian States and each of the external territories. The external territories comprised those independent territories in Australia, as well as islands that, despite belonging to one of the state’s jurisdictions, are located far away and therefore have a different flora than the one in the state they are associated with. As such, the external territories included Ashmore Reef (AR), Cartier Island (CaI), Christmas Island (ChI), Cocos Islands (CoI), Coral Sea Islands (CSI), Heard Island (HI), Lord Howe Island (LHI), MacDonald Island (MDI) and Norfolk Island (NI). Although, external territories were excluded from the state-by-state comparison explained below, we still provided, for each of them, the information on the present flora extracted from the APC.

For those species that had a combination of statuses within the same state, we condensed the information into one introduction status per species and state or territory (see Tables 1–3 for details of status conversion). This happened for 0.3% of the vascular plant species records at the national scale. Species recorded as “native and naturalised” or “native and naturalised and uncertain origin” within a certain state were reassigned as ‘native colonising’, as they are native to part of the state, but colonising and therefore considered introduced in other areas within the same state. Similarly, species recorded as “native and doubtfully naturalised” or “native and uncertain origin” were reassigned to ‘native potentially colonising’ (see Martín-Forés et al.22 for discussion on terminology in invasion ecology across Australia).

State-by-state curation process

For the eight repositories at the state scale, we developed standardised datasets (see Martín-Forés et al. 2023). To do so, we matched each species canonical name (i.e., species name without the author) displayed in the state plant census with the APC to obtain the equivalent accepted name from the APC for those species that were known synonyms or had been misapplied. We then developed a conversion procedure to standardise all introduction statuses across Australian states (see Tables 1–3 for details). Standardised nomenclature adapted from the unified framework on biological invasions8 was used to combine the information about presence (present/extinct), origin (native/introduced/uncertain) and introduction status (casual/ naturalised/ invasive) for all data sources. For example, for a species present, introduced, and for which its degree of establishment according to Darwin Core was known as ‘established’, we recorded it directly as naturalised. To facilitate understanding of the different terminology when used at both state and national level, we have provided a glossary with specific meanings for each term at both scales and according to different sources of vocabulary for invasion ecology (Tables 4, 5). Details for rationale behind this decision and discussion on equivalences between different frameworks and standards can be found in Martín-Forés et al.22.

Curation process for the Australian GRIIS

The Australian GRIIS is structured in three different datasets that were downloaded from (https://cloud.gbif.org/griis/resource?r=griis-australia; GRIIS v1.9, accessed February 2023). The three datasets were merged into one and only the records for plants were kept. The taxa that appeared recorded as ‘invasive’ based on impact were changed to ‘harmful invasive’ according to the definitions provided in detail by Essl et al.26 and Martín-Forés et al.22.

Matching introduction statuses between the state censuses and the APC (step 3)

In each of the datasets at the state level, we matched the corresponding accepted names in APC to the complete APC dataset to obtain the corresponding state introduction status in the APC (Fig. 1).

We also matched the taxon status to identify the species that have been excluded from the APC, and those that were recorded in the APC as pro-parte or pro-parte misapplied, so that we could treat these cases individually. Pro-parte or pro-parte misapplied species are, respectively, those that can be both an accepted name and synonym for another accepted name, or that correspond to two different accepted names, and those that have been mis-assigned to different accepted species. If pro-parte and pro-parte misapplied species only matched one accepted species from the APC, we kept those records. However, in cases where they matched two or more accepted species from the APC, we stated that it was impossible to elucidate to which one the species name appearing on the state census would correspond.

Once matched, the distribution and introduction status within each Australian state is scored by cross-checking extracted information from the APC and the state flora censuses accessed from the state herbaria. As such, we compared species scientific name (including authorship), canonical name, and introduction status recorded on the state census with those recorded on the APC (See supplemental Table S1 in Supplementary material for details). This comparison between the state plant censuses and the distribution information recorded on the APC detected potential mismatches (see Martín-Forés et al.22 for discussion on mismatches in introduction status at the state and territory levels in Australia).

Assigning a unified status for a given species in each state (step 4)

The script we created easily identified and tackled mismatches in nomenclature, introduction status or even presence within a certain area (Fig. 1). Where introduction statuses differed between APC and the state plant census, we applied the prioritisation procedure proposed in the integration exercise conducted by Martín-Forés et al.22 to address inconsistencies regarding introduction status at the jurisdictional level in Australia. According to such prioritisation procedure, the introduction status that has advanced the furthest along the continuum prevails. This was decided as a precautionary measure for addressing potential invasion and to be as conservative as possible when designing management and eradication programs. As such, when a species was not listed on the APC or was recorded on the APC as not present in a given state, we kept the introduction status recorded in the state plant census. For species that appeared in both sources but displayed a mismatch in the introduction status, we prioritise the most conservative introduction status (i.e. the one advanced the furthest along the continuum22). In all component datasets developed at the state level, we incorporated a new column with the unified status for each species in such state (See Figshare to access all the standardised state-by-state datasets for all Australian jurisdictions27).

Integrating unified state statuses into a national status and cross-reference with the GRIIS (step 5)

The unified status for each species across all Australian states were combined and integrated into a status at the national level (Fig. 1). For each accepted species on the APC dataset, we matched the unified status obtained for each state in each of the standardised datasets at the jurisdictional level. When the species did not appear listed on the state census, we kept the introduction status recorded on the APC; when the species was not originally listed on the APC, but only on the state census, we did not incorporate it at national level, therefore appearing only on the specific standardised datasets for the corresponding jurisdiction (Figshare27). Finally, when the species was recorded in both sources, we integrated the introduction statuses among states into an introduction status at the national scale. To conduct such integration at the national scale, we followed the prioritisation procedure described by Martín-Forés et al.22 to address inconsistencies regarding introduction status at the national level. The prioritisation procedure at the national scale differed from the one used at the state level as follows: if a species was native to at least one state, it was considered native at the national scale. If it was not ‘native’ to any state, but it was recorded as ‘native colonising’ (or native potentially colonising) in any state, it was considered as native colonising at the national scale. If a species was neither native in any possible form to any state, nor having uncertain origin, then the introduction status that had advanced the furthest along the continuum was the one prevailing at the national scale. Only if the species was not present in any state was it then recorded as presumed extinct at the national scale (see Martín-Forés et al.22 for detailed prioritisation procedure at the national scale).

We then cross-referenced the unified introduction status we obtained at the national scale with the one appearing on the Australian GRIIS21 and harmonised the information recorded in both into the AFA status at national scale (See Supplementary material for details).

Assigning a unified status (AFA status) for a given species at the national scale (step 6)

Where introduction status differed between the national status and the GRIIS, we followed the steps proposed by Martín-Forés et al.22 (Fig. 1).

For species for which the introduction status resulting from combining state data sources at the national scale was alien in any form, and for which the status recorded on the GRIIS was ‘invasive’, the unified AFA status was stated as ‘harmful invasive’. This was due to the fact that the GRIIS classifies invasive species based on negative impact, not only on distribution along the continuum (see Martín-Forés et al.22 for detailed discussion). When other mismatches were identified (e.g., species that are native to at least one Australian state but appeared recorded as introduced or invasive (i.e., harmful invasive) in the GRIIS), the AFA status corresponded to the one resulting from integrating the state statuses at the national scale as detailed in the previous step.

Depositing the R script, standardised state-by-state datasets and the AFA dataset in open repositories (step 7)

Finally, we made sure the R script was accessible both via Figshare and github to facilitate updated versions of the AFA with new releases of the state censuses or the APC (Fig. 1).

The standardised datasets at the state level are accessible for informative purposes or in cases where jurisdictional authorities can make use of the information provided there.

The AFA dataset contains a comprehensive up-to-date information on alien flora at the national level.

Data Records

The dataset is available at Figshare, https://doi.org/10.6084/m9.figshare.2351347827. The data records are structured as follows:

The AFA national dataset is available for direct download in a.csv file. The AFA national dataset contains information on taxa canonical and scientific names, authorship, taxonomic rank (e.g. species epithet, subspecies epithet), and the taxon status according to the APC (which in all cases is accepted for this dataset), the unique ID status for each taxon is also provided. Moreover, there are columns with the unified introduction status for all taxa in each state and main territory, and with the introduction statuses populated from the APC for the external territories. Afterwards, there is the national introduction status resulting from the integration at the national scale, the GRIIS status and a column resulting from the comparison between the two. Finally, there is the AFA unified status that corresponds to the unified status for each taxon across all data sources at the national scale. In addition, the AFA unified status has been converted into its corresponding Darwin Core equivalences for the terms establishment means and degree of establishment.

We provide four more folders in Figshare27. The first folder is a zip folder called AFA_state-by-state datasets that contains all the state-by-state standardised datasets (eight in total) in.csv format which were produced as intermediate steps for creating the AFA national dataset. The state-by-state standardised datasets are all displayed following the same structure, including information on the taxonomy recognised by the state censuses and the APC, the introduction status recorded on both the state census and the APC, the taxon status according to APC (e.g. accepted, misapplied, excluded), a column resulting from the comparison of introduction status between the state census and the APC, and a column with the unified introduction status at the state level. In addition, the unified introduction status at the state level was converted into its corresponding equivalences from the Darwin Core, establishment means and degree of establishment terms.

The second zip folder is called AFA_metadata and contains two.csv with the metadata for the AFA national dataset and the state-by-state datasets.

The third folder is a zip folder called AFA_R_Scripts that contains three R files which were produced as intermediate steps for creating the AFA national dataset. The AFA_main_R_script contains the condensed information on how to reproduce the process to create the AFA dataset. In such instructions, we duplicated the first step to allow for users to either replicate the static copy or the AFA presented here, or alternatively to create updated versions in the future with new releases of the APC and the state censuses. In addition, there are two more R files containing the primary and secondary functions to be used in the main script. These two files containing the functions can be easily loaded following the steps detailed on the main R script. We split the files, so that in the future, the script developed to create this dataset can be easily adapted as needed and reused to create updated versions of the AFA with new releases of the state and main territory plant censuses, and of the Australian Plant Census. The three R files are also accessible via github (https://github.com/MartinFores/AFA).

Finally, the fourth zip folder is called AFA_raw_data and it contains the raw versions obtained from the different data sources that were employed to create the AFA national dataset presented here, together with a pdf file outlining the corresponding citations for each data source. As such, the users can reproduce the AFA national dataset presented here, by accessing a static copy of the APC, the three datasets conforming the GRIIS, and static copies of the state and main territory plant censuses for the corresponding versions detailed on Tables 1–3.

Technical Validation

This standardised system is itself the result of a technical validation. By cross-referencing all up-to-date official sources of vascular plant censuses in each of the Australian states, main territories and at the national scale, we have identified all the existing mismatches to date, and when a mismatch has been detected, the tool ensures that the most conservative approach is followed.

When the metadata of certain state census appeared unclear, we contacted the relevant authority to ensure accuracy in the interpretation of the terminology employed and subsequently in our conversion.

Additionally, we consulted with a subcommittee of the Council of Heads of Australasian Herbaria (CHAH), the Herbarium Information Systems Committee (HISCOM; https://chah.gov.au/the-herbarium-information-systems-committee/), and each of the data custodians for each state. We took this step to ensure the conversion of introduction statuses was accurate. As a result of this process, we obtained feedback to incorporate the Darwin Core terminology in our standardised system to also be inclusive of other international standards that may be implemented in the foreseeable future.