Ornamental Decision Support Tool Final Report for the ‘Plant Sure Environmentally Safe Ornamental Plant Scheme Plant’ project

Nola Hancock, Victoria Graham, Vanessa Adams, Tim Maher, Anthony Manea and Michelle Leishman Department of Biological Sciences, Macquarie University

22 March 2018

Please cite this publication as:

Hancock, N., Graham, V., Adams, V., Maher, T., Manea, A. and Leishman, M. (2018). Ornamental Plant Decision Support Tool Final Report for the ‘Plant Sure Environmentally Safe Ornamental Plant Scheme Plant’ project. Macquarie University, Sydney

For further correspondence contact: [email protected]

i

Table of Contents

Table of Contents ...... ii List of Figures ...... iii List of Tables ...... iii Acronyms Used in This Report ...... iv Acknowledgements ...... iv Executive Summary ...... v Introduction ...... 1 Weed Risk Assessment Tools ...... 2 Scope of project ...... 2 Objectives ...... 3 Methods ...... 3 Identification of plant risk assessment tools ...... 3 Design & development of the plant risk assessment tool ...... 4 Invasive listing - scientific name ...... 6 Prohibited invasive plants ...... 6 Restricted invasive plants ...... 6 Weed species of Western Australia ...... 6 Testing the accuracy of the risk assessment tool ...... 9 Results & Recommendations ...... 12 Results of the OPDST accuracy testing ...... 12 Recommendations ...... 18 References...... 20 Appendix 1...... 22 Appendix 2...... 23 Appendix 3...... 27 Appendix 4...... 28

ii

List of Figures

Figure 1: Flow chart showing the screening process used to generate the list of weed risk assessment tools that were assessed against the review criteria...... 4

Figure 2. Frequency histogram of the scores of the 58 species evaluated in the trial of the Plant Sure Ornamental Plant Decision Support Tool divided into categories of invasive or non-invasive to assess accuracy of the tool. Invasive species are those that have been reported as being invasive anywhere globally. Non-invasive represents species not reported as invasive anywhere globally...... 12

Figure 3. The scores of the 58 species evaluated in the trial of the Plant Sure Ornamental Plant Decision Support Tool, categorised in NSW as invasive, naturalised and non-naturalised ...... 13

Figure 4. Scatter plot showing the correlation between the pre-screen score and the main screen score for the 58 plants assessed in the trial of the Plant Sure Ornamental Plant Decision Support Tool...... 14

List of Tables

Table 1. Commonwealth and State weed lists of noxious weeds and prohibited plants .. 6

Table 2: List of the 58 species evaluated in the trial assessment...... 10

Table 3. Comparison of the pre-screen and main screen results...... 14

Table 4. Response rates of the 24 questions included in the Plant Sure Ornamental Plant Decision Support Tool...... 15

Table 5. Example of a risk matrix to be completed after the evaluation process ...... 17

iii

Acronyms Used in This Report

OPDST Ornamental Plant Decision Support Tool

NSW New South Wales

RFQ Request for Quotation

PRE Plant Right Evaluation Tool

WRA Weed Risk Assessment

Acknowledgements

This research was funded by the Environmental Trust via a grant administered by a consortium led by the Nursery & Garden Industry NSW & ACT (NGINA) and by the NSW Office of Environment and Heritage.

iv

Executive Summary

Invasive species incur significant costs to natural environments and economies. The latest Australia State of the Environment Report (2016) describes the impact of invasive species as undiminishing and increasing when combined with other stressors (Cresswell and Murphy 2017). Horticulture accounts for a substantial proportion of the plant species introduced into Australia (Gallagher & Leishman 2015), and the horticultural industry is the source of the majority of exotics that have become invasive. Whilst strict importation controls currently regulate the arrival of new exotic plant species, the sale of existing exotics is not regulated and there is the potential that some of these will become garden ‘escapees’ and invasive.

The Plant Sure Environmentally Safe Ornamental Plant Scheme project is a collaborative undertaking that aims to limit the supply or discourage the use of non- native plants with high potential for invasiveness. This report introduces the Plant Sure Decision Support Tool (OPDST) and follows on from Maher et al (2017) wherein the development process of the OPDST is discussed.

The Plant Right Evaluation Tool (PRE) (Conser et al 2015) was identified as the most suitable template for the Plant Sure OPDST. Several modifications were made to the PRE to adjust for Australian conditions and to fit the requirements of the Plant Sure project. The key modifications were:

I. A pre-screen assessment was added to circumvent a full assessment if the plant is listed on any Commonwealth, State or Territory prohibited plant list or if it scores 8 or greater in the pre-screen questions;

II. The inclusion of questions that assess species against the bioclimatic zones of the target region (NSW);

III. A cultivar tool was added as an extension. The first question in the main tool draws out cultivars to be evaluated using the cultivar-specific tool;

IV. Consideration of hybridisation potential of the species based on available evidence of other species in the hybridising with invasive species in the target region; and

V. An assessment to identify if the area of climate suitability for the species is projected to increase under climate change.

The tool was designed to balance the amount of time it takes to complete (i.e. the number of questions) and the statistical ability to predict invasiveness. For example, the final set of questions only retained plant biological questions (e.g. seed size and plant form) that contributed most to statistical prediction.

In summary, the OPDST is a series of 24 questions that evaluates and accepts or rejects species according to their risk of invasiveness. The OPDST achieved 100% accuracy in evaluating 58 trial species; either rejecting them as invasive or accepting them as non-invasive, excluding four species requiring further evaluation. There was a

v

high average response rate for all questions (90%), and on average it took 115 minutes- to complete one evaluation.

Whilst the tool is easy to use and can be adapted to any scheme environment, its effectiveness is limited by the quality and quantity of data that is sourced and available. It is therefore recommended that assessments are undertaken and verified by a panel of experts and that completed assessments are made available to other end users, preferably on-line.

vi

Introduction

Weeds pose a multi-billion dollar cost to the Australian economy (Sinden et al., 2004) and present a significant threat to biodiversity (Coutts-Smith and Downey, 2006). Australia utilises rigorous processes that prohibit plant species from entering the country unless they undergo a formal risk assessment and / or appear on the Permitted Seeds List (ISC, 2009). This process has prevented the importation of approximately 1,500 high-risk exotic species since its inception in 1998 (ISC, 2009). Several states of Australia use post-border assessment frameworks to prioritise the management and control of weed species already present (Virtue, 2008; Weiss and McLaren, 2002). The Office of Gene Technology Regulator (http://www.ogtr.gov.au/) uses pre-border assessment frameworks to evaluate and control species not yet present, including the importation of genetically modified plants.

However, unless banned at the State or Territory level, the sale of those species that were imported prior to 1998 is generally not restricted (ISC, 2009). Estimates of the exotic plant species introduced into Australia generally range from 26,000-29,000 (Scott et al 2014; Gallagher & Leishman, 2015). However, the latest State Of the Environment report (2016) lists the number of introduced plant species at 41,000, considerably more than previous estimates but this data was derived from an unpublished state government source (Metcalfe & Bui, 2017). Of the 26,000 – 29,000 exotics, approximately 90% were introduced for ornamental or horticultural purposes (ISC, 2009; Gallagher & Leishman, 2015), approximately 7% have become weeds in native vegetation (Gallagher & Leishman, 2015), approximately 10% have become naturalised (Scott et al 2014; Dodd et al 2015; Gallagher & Leishman, 2015) and the annual rate of further naturalisation is estimated at 12% (Scott et al 2014). Furthermore, 23% of the exotic species that have not naturalised are known to be invasive outside of Australia (ISC, 2009). These data suggest that it is highly likely that more exotic species will become invasive in Australia. Thus, a cost-effective way to restrict the spread of potentially invasive plants is to intercept them at the nursery level before they ‘jump the garden fence’.

The Plant Sure Consortia, led by the Nursery & Garden Industry of NSW & ACT, the Australian Institute of Horticulture, the NSW Office of Environment and Heritage, and NSW Department of Primary Industries, developed a proposal for an environmentally- safe ornamental plant scheme in NSW, Australia. The Plant Sure Environmentally Safe Ornamental Plant Scheme project is a two-phased collaborative initiative that aims to limit the supply or discourage the use of non-native plants with high potential invasion risk. The first phase of the project is to develop a voluntary accreditation or certification scheme to engage relevant industries in promoting environmentally safe plants and the removal or avoidance of the use of exotic plants with a high environmental risk. This phase has two components: (1) develop a ‘decision support tool’ (a plant risk assessment tool) that allows plants to be assessed and categorised based on their level of environmental weed risk; and (2) design a model scheme for the project.

1

This report, commissioned by NGINA, and managed by the consortia with guidance from other stakeholders via reference groups, focuses on the delivery of component (1), the decision support tool, hereafter referred to as the Ornamental Plant Decision Support Tool (OPDST).

Weed Risk Assessment Tools

The development of the OPDST was based on a commonly-used and effective tool that identifies plants with a high likelihood of becoming invasive with the aim to restrict their introduction. This tool is generally referred to as a weed risk assessment (WRA) tool.

A WRA utilises information on a species’ current international weed status, environmental requirements (including climate) and relevant biological attributes to predict its invasiveness (Pheloung et al 1999).The first WRA was developed in Australia in the late 1990’s to assess the potential of proposed imported taxa of becoming agricultural or environmental weeds (Pheloung et al 1999). This WRA comprised 49 questions to assess and categorise taxa as weeds (reject), non-weeds (accept) or more information needed to evaluate (evaluate further). When tested against known weeds and non-weeds, 84% of serious weeds and most minor weeds were rejected, 93% of non-weeds were accepted and 29% were in the ‘evaluate further’ category (Pheloung et al 1999). The challenge for WRAs is to not reject species with a low risk of invasiveness (false positives) or accept species with a high risk of invasiveness (false negatives; Conser et al 2015). Over the last 30 years, WRA methodologies have significantly improved and now boast higher levels of accuracy in predicting invasiveness and non- invasiveness. Examples include the Plant Right Evaluation Tool (hereafter referred to as the PRE; Conser et al 2015) and the UF/IFAS Assessment of Non-native Plants in Florida’s Natural Areas (Lieurance et al 2013). Scope of project

The purpose of this project was to inform the development of a simple, easy to use screening tool (WRA) that can be used voluntarily by anyone in the green life industry (including growers, retailers, wholesalers and landscapers) to assess the environmental risk of ornamental plants available for sale. Ideally, the tool should cover both aquatic and terrestrial plants, where possible using the same methodology. It should focus on invasiveness, especially ecological impacts, but include other agricultural and social impacts (e.g. health) where possible. The geographic scope is NSW but the tool should be national-ready in that it can be readily rolled-out across Australia and should be sector-ready so that it can be adopted by any range of sectors or industries (e.g. not specific to the horticultural industry). The tool will be the underlying framework used to design an accreditation scheme.

2

Objectives

The objective of this report is to detail the recommended screening tool, as described in Maher et al (2017), to be used as a prototype for the Plant Sure OPDST, and in particular to;

I. Describe the modifications made to the recommended screening tool; II. Test the modified tool using Australian plant species; and III. Provide recommendations on the application of the tool for the Plant Sure scheme.

Methods

Identification of plant risk assessment tools

A review of existing national and international decision support tools assessing weed risk of plants (including known weeds and plants not yet known to be weedy) was undertaken to identify the most suitable tool for the Plant Sure scheme. Full details of the process are detailed in Maher et al (2017) and a brief summary is provided below.

The papers, tools and processes outlined in the Plant Sure RFQ were reviewed and 57 tools/processes were identified from our literature search (Figure 1). Of the 57 tools, only those that presented the development of a new WRA model or represented a significant improvement on previous models were included to avoid duplication. Models used for pest risk assessment on a range of taxa were also included if they met enough of the criteria and could be adapted for the OPDST. Consequently, 13 tools that best met the selection criteria were selected for thorough evaluation. The screening process ensured that all reviewed tools met the first RFQ criteria of reflecting best practice science and were publicly available.

3

Tools identified from initial search (n =57)

Identification

Tools after duplicate protocols removed

(n = 37)

Screening Tools after screening Tools excluded due to (n =23) ineligibility (n =10)

Tools included in the review

Included (n = 13)

Figure 1: Flow chart showing the screening process used to generate the list of weed risk assessment tools that were assessed against the review criteria.

Design & development of the plant risk assessment tool

The review identified the PRE (https://plantright.org/about-invasive-plants/plant-list/) as most suitable as a model tool for the Plant Sure project. The PRE questions, scoring mechanism and thresholds for risk categories were retained, with some modifications made to fit the specific purpose of the Plant Sure scheme. The final questions and scoring mechanism for the Plant Sure OPDST are detailed in Appendix 1.

The following key modifications were made to the PRE to develop an ornamental plant risk assessment tool that fits the Plant Sure brief:

I. A pre-screen assessment was added to circumvent a full assessment if the plant is listed on any Commonwealth, State or Territory prohibited plant list (Q.2); or if it scores 8 or greater in the pre-screen questions 3 - 8;

4

II. Questions to assess species against the bioclimatic zones of the target region (NSW) were added (Qs 3, 5, 10 & 11);

III. A cultivar tool was added as an extension. The first question in the main tool (Q.1) draws out cultivars to be evaluated using the cultivar-specific tool;

IV. Consideration of hybridisation potential of the species based on available evidence of other species in the genus hybridising with invasive species in the target region was added (Q. 24); and

V. An assessment to identify if the area of climate suitability of the species is projected to increase under climate change was added (Q. 23).

Further descriptions of these modifications and the specific criteria from the Plant Sure brief that they address are detailed below:

I. Contain an initial risk assessment ‘screen’ that can quickly reject species that are documented weeds without undertaking a full assessment.

An initial screen to rule out plants that are already present on a legislated weed list was not present in the PRE. This was addressed by incorporating the initial pre-screen question from the UF/IFAS Assessment of Non-native Plants in Florida’s Natural Areas (Lieurance et al 2013): “Is this species or cultivar listed on any Commonwealth or state noxious or prohibited plant lists? If yes, stop assessment.” We selected the following authoritative lists with guidance from the consortia (Table 1). It includes two Commonwealth lists and eight State lists. This question appears as question number 2 in the Plant Sure OPDST. This list is available to users of the OPDST in the ‘Web sites’ work sheet of the OPDST (Appendix 1).

The PRE took the top six questions of the complete tool that best predicted weeds as a rapid pre-screen to quickly reject high risk plants without undergoing the full assessment. We added these six questions at the beginning of the OPDST to serve as a rapid pre-screen (questions 3 – 8). The pre-screen instantly rejects species that have become invasive in other places with a similar climate to the target region and that have high reproductive potential and/or that smother native plants. If a plant scores 8 or greater in the pre-screen questions 3 – 8 then it is rejected without undergoing the full assessment.

5

Table 1. Commonwealth and State weed lists of noxious weeds and prohibited plants

Jurisdiction List name Link

Commonwealth •Weeds of http://weeds.dpi.nsw.gov.au/WeedListPublics/C National ategoryResults?showImages=True&categoryId= Significance 5&pageTitle=National%20Environmental%20Al ert%20List%20weeds •National Environmental http://weeds.dpi.nsw.gov.au/WeedListPublics/C Alert Weed list ategoryResults?showImages=True&categoryId= 5&pageTitle=National%20Environmental%20Al ert%20List%20weeds

NSW NSW Weedwise http://weeds.dpi.nsw.gov.au/

VIC Invasive plants http://vro.agriculture.vic.gov.au/dpi/vro/vrosite.n listing - scientific sf/pages/weeds_listing2_a name N.B. this is a complete list of weed species in Vic. The noxious weeds are indicated by (Nox)

SA Consolidated list http://www.pir.sa.gov.au/__data/assets/pdf_file/ of declarations of 0003/231924/Declaration_of_Animals_and_Pla animals and nts_-_July_2017.pdf plants July 2017 QLD Prohibited https://www.business.qld.gov.au/industries/farm invasive plants s-fishing-forestry/agriculture/land- management/health-pests-weeds- diseases/weeds-diseases/invasive- plants/prohibited

https://www.business.qld.gov.au/industries/farm Restricted s-fishing-forestry/agriculture/land- invasive plants management/health-pests-weeds- diseases/weeds-diseases/invasive- plants/restricted

WA Weed species of https://florabase.dpaw.wa.gov.au/weeds/ Western Australia

TAS Invasive species http://dpipwe.tas.gov.au/invasive- – weed index species/weeds/weeds-index/weeds-index- alphabetic-by-scientific-name ACT Pest plants and http://www.legislation.act.gov.au/di/2015- animals (pest 59/current/pdf/2015-59.pdf plants)

6

Declaration 2015 (No 1) NT Declared weeds https://nt.gov.au/__data/assets/pdf_file/0016/25 in the Northern 2133/declared-weeds-in-the-nt.pdf Territory

II. Allow assessment of species in a variety of bioclimatic zones

The PRE allowed the risk of an ornamental plant to be assessed separately for each of five bioclimatic zones in California by using a purpose-built, online climate-matching tool. A literature search was conducted to explore a number of approaches in order to answer the climate related questions in the PRE for NSW. In particular Eyre et al (2012) provided an extensive review of methods for mapping the suitability of the climate for pest risk analysis in order to act as a guide for pest risk assessors. In reviewing these different methods, most proved to be very time consuming, data intensive and required a specialised skillset. However, Eyre et al (2013) suggest a more simplistic approach and short cuts for assessing risk of a species across different bioclimatic zones in the target region. While this approach is not as accurate and carries more uncertainty, it is still informative and much less time consuming than other methods we reviewed. Therefore, in order to keep with the user-friendly objective of the project, we decided to adopt a similar approach to Eyre et al (2013) in addressing the climate-related questions in the OPDST. Our method uses the freely-available online climate-matching model, Climatch (http://data.daff.gov.au:8080/Climatch/climatch.jsp) developed by the Department of Agriculture and Water Resources. Climatch allows users to identify the similarity in the climate between a species’ existing range and other regions in Australia. Climatch has previously been used for this purpose in Australia to answer the pre-border WRA and post-border risk assessments conducted by Western Australia’s Department of Agriculture and Food (Brown et al 2006). We have created a set of instructions on how to apply Climatch in answering the climate-related questions in the OPDST (Appendix 2). Climatch is used in question numbers 3,5,10 & 11 in the Plant Sure OPDST.

III. Allow assessment of hybrids, varieties, cultivars and/or intraspecific variation (or where this is not possible contain a mechanism to recognise these taxa and identify the information needed for future assessment)

The PRE was used to test many plant varieties and cultivars independently of the parent species. An alternative to this is a more detailed separate protocol designed especially for assessing cultivars (Appendix 3). This approach might better distinguish low risk cultivars from the parent species, but finding the information needed to complete an assessment is difficult. For example, one question used in the New York State Cultivar Invasiveness Assessment Protocol (Jordan et al 2012) is: “How likely is this cultivar to establish expanding populations?” Such information will not be available for most cultivars, especially if they are new to the market. It is for this reason that we decided to trial both the main screening tool (PRE) and a cultivar tool (New York State Cultivar

7

Invasiveness Assessment Protocol) and see how readily information was available to answer each set of questions, and if cultivars were categorised differently by the two tools. Cultivars that lack information to answer the questions are put into an ‘evaluate further’ category until more information is available, which may require contact with the developer of the cultivar or field trials to test how the cultivar differs from the parent species. This question appears as question number 1 in the Plant Sure OPDST which subsequently directs the user to the New York State Cultivar Invasiveness Assessment Protocol.

IV. Include methods to assess risks associated with hybridisation, genetic pollution and taxonomic/genetic relatedness (including via conventional breeding and natural processes) and risk of genetically modified material

The draft PRE originally considered hybridisation risk but this was later removed because it was not a good predictor of whether a species was invasive. The Plant Sure project reference group recommended that information on hybridisation would be useful to certain stakeholders. Therefore, a question was added to the Plant Sure OPDST as follows: “Is there evidence of a species in this genus hybridising with a species that is invasive in the target region?” This question assesses whether there is previous evidence of a closely related species forming an invasive hybrid with a species that is already present in the target region. This question serves as a warning and does not contribute to the score. This question appears as question number 24 in the Plant Sure OPDST.

The risk of genetically modified material was also not included in the PRE tool, because of the rigorous pre-border assessment process that already exists for genetically modified organisms in Australia (http://ogtr.gov.au/internet/ogtr/publishing.nsf/Content/import-reqs-1).

V. Consider how suitable habitat for the species may expand, contract or shift under future climates

We added a question on how climate projections may affect the establishment and spread of a given species in the target region. The question is answered using the online Weed Futures tool (http://www.weedfutures.net/) developed by Macquarie University, which has modelled the risk of establishment and spread of over 800 invasive and naturalised species under future climate projections. Until this resource has been updated to include all introduced species, this question should serve as a warning and not contribute to the score. This question “Is the distribution of this species projected to increase under future climate change in the target region?” is number 23 in the Plant Sure OPDST.

8

Testing the accuracy of the risk assessment tool

We screened 58 plant species to test the accuracy of the tool. Of the screened plants, 16 were known as invasive (28%), 18 as naturalised (31%) and 24 as non-naturalised (41%). We selected species for the trial that represented both invasive and non-invasive species without adding undue bias to our sample. Base-rate neglect is a documented weakness in many existing week risk assessments, where accuracy trials do not reflect weed prevalence (Hulme, 2012). To decrease bias in our sample, we aimed to bring the plant ratio of invasive / non-invasive as close as possible to the base rate of weeds, which is explained by the ‘tens rule’. The ‘tens rule’ states that of every 1,000 species introduced, roughly 100 are likely to naturalise and roughly 10 are likely to become invasive. Our initial testing of 49 species adhered to this ratio but a subsequent trial of a further 9 species altered the ratio in favour of a greater ratio of invasive species. To rectify this, further testing of naturalised and non-naturalised species is recommended.

In addition to reflecting the base-rate of invasive plants, we aimed to capture species that are commonly traded by plant retailers, by searching for species on the Evergeen Connect top sellers list (https://www.evergreenconnect.com.au/index.php). We also collated species’ nominations from members of the consortia to capture any species of particular interest or concern. From the nomination list, we selected species from the families with the most weed species in Australia using the Weed Futures database (http://weedfutures.net/), while also representing a diverse range of growth forms (tree, , herb, vine and grass). For the naturalised and non-naturalised categories, we prioritised species that had been present in Australia for greater than 50 years and aimed to represent a wide range of taxonomic families and growth forms in our sample (Table 2). We tested three cultivars using both the main screening tool and the cultivar screening tool, which is based on the New York State Cultivar Invasiveness Assessment Protocol (http://nyis.info/?action=israt; Appendix 3).

For the trial, we evaluated the risk of plants becoming invasive in NSW, but the OPDST can be applied to any target region. We evaluated each plant using the main scoring assessment and the rapid pre-scoring assessment by summing the points allocated for each question (Appendix 1). As the PRE had determined which questions contributed to the predictability of invasiveness and non-invasiveness using a two-tailed Fischer’s Exact Test (Conser et al 2015), we did not repeat this analysis to assess the predictive power of each question. We measured the response rate for each question to assess the difficulty in answering the question.

9

Table 2: List of the 58 species evaluated in the trial assessment.

Family Species name Common Growth Main Outcome name form screen score Invasive in NSW Cyperaceae Cyperus esculentus Yellow Herb 14 Evaluate Nutgrass further Adoxaceae Viburnum odoratissimum Emerald lustre Shrub 17 Reject var. awabuki* Amygdalaceae Prunus cerasifera Cherry plum Tree 16 Reject Amygdalaceae Prunus persica Peach Tree 15 Evaluate further Asparagaceae Asparagus aethiopicus Asparagus fern Shrub 21 Reject Gazania rigens Coastal Herb 20 Reject gazania Asteraceae Sphagneticola trilobata Creeping Ox- Herb 18 Reject eye Balsaminaceae Impatiens walleriana Balsam Herb 19 Reject Euphorbiaceae Triadica sebifera syn. Chinese tallow Tree 22 Reject Sapium sebiferum Fabaceae Robinia pseudoacacia Robinia Tree 22 Reject Malaceae Rhaphiolepis indica Indian Shrub 17 Reject hawthorn Onagraceae Gaura lindheimeri Butterfly bush Herb 17 Reject Poaceae Miscanthus sinensis Chinese fairy Grass 21 Reject grass Poaceae Cenchrus setaceus Fountain Grass Grass 23 Reject Solanum seaforthianum Brazilian Vine 17 Reject Andrews nightshade Verbenaceae Duranta erecta Geisha girl Shrub 22 Reject Average (invasive) 19 Naturalised in NSW Adoxaceae Viburnum tinus Laurustinus Shrub 15 Evaluate further Asteraceae sordida Purple torch Shrub 18 Reject Betulaceae Betula Pendula Silver birch Tree 16 Reject Bignoniaceae Handroanthus Yellow trumpet Tree 11 Accept chrysotrichus tree Caryophyllaceae Dianthus barbatus Sweet william Herb 6 Accept Commelinaceae Dichorisandra thyrsiflora Blue ginger Herb 11 Accept Cycadeaceae Cycas revoluta Sago palm Shrub 8 Accept Ericaceae Rhododendron ponticum Common Shrub 17 Reject rhododendron Euphorbiaceae Euphorbia tirucali Pencil tree Tree 8 Accept Fabaceae Senna multiglandulosa Glandular Shrub 7 Accept senna Hemerocallidaceae Phormium tenax New Zealand Herb 17 Reject flax Malvaceae Hibiscus rosa-sinensis Chinese Shrub 6 Accept hibiscus Oleaceae Fraxinus excelsior European ash Tree 17 Reject Poaceae Alopecurus pratensis Golden foxtail Grass 20 Reject grass Poaceae Pogonatherum paniceum Baby panda Grass 2 Accept grass Rutaceae Murraya paniculata Mock orange Shrub 8 Accept Solanaceae Lycium barbarum Chinese Shrub 17 Reject boxthorn Strelitziaceae Strelitzia reginae Bird of paradise Shrub 7 Accept

10

Family Species name Common Growth Main Outcome name form screen score Average (naturalised) 12 Non-naturalised in NSW Adoxaceae Viburnum plicatum f. Doublefile Shrub 7 Accept tomentosum viburnum Amaryllidaceae Lycoris aurea Golden spider Herb 7 Accept lily Apocynaceae Ceropegia woodii Chain of hearts Vine 4 Accept

Araceae Aglaonema nitidum Chinese Herb 6 Accept evergreen Asparagaceae Liriope muscari Big blue lilyturf Herb 7 Accept Asparagaceae Ophiopogon planiscapus Black mondo Herb 7 Accept 'Nigrescens'* grass Asteraceae Chrysanthemum pacificum Pacific Herb 8 Accept chrysanthemum Asteraceae Cosmos sulphureus Yellow cosmos Herb 15 Evaluate further Cactaceae Lophophora williamsii Peyote cactus Herb 4 Accept Ericaceae Rhododendron indicum Alba magnifica Shrub 9 Accept Euphorbiaceae Codiaeum variegatum Garden croton Shrub 1 Accept Fabaceae latifolia Yesterday Shrub 6 Accept today and tomorrow Fabaceae Bauhinia variegata Orchid tree Tree 19 Reject Iridacaeae Watsonia hysterantha Autumn Herb 4 Accept watsonia Iridaceae Iris chrysographes Black iris Herb 3 Accept Magnoliaceae Magnolia grandiflora Little gem Tree 10 Accept Magnoliaceae Michelia figo Port wine Shrub 2 Accept magnolia Plantaginaceae Hebe buxiflora Box hebe Shrub 0 Accept Poaceae Saccharum officinarum Purple sugar Grass 3 Accept var. violaceum* cane Polygalaceae Polygala dalmaisiana Sweet pea Shrub 3 Accept shrub Rosaceae Pyrus calleryana Callery pear Tree 18 Reject Rosaceae Photinia davidiana Chinese Shrub 5 Accept photinia Rubiaceae Gardenia augusta Gardenia Shrub 5 Accept Theaceae Camellia japonica Common Shrub 3 Accept camellia Average (non- 7 naturalised) Average (all species) 12

*cultivars

11

Results & Recommendations

Results of the OPDST accuracy testing

The final list of the 58 species evaluated in the trial (Table 2) comprised 16 (28%) known invasive plants in NSW, 18 (31%) naturalised plants in NSW and 24 (41%) non- naturalised plants in NSW. The final score for the trial species is the sum of the 24 OPDST questions (out of a possible 25 points). Species with a score lower than 13 are accepted as low invasive risk, between 13 – 15 need further evaluation and species with a score above 15 are rejected due to high invasive risk. The cumulative scores for the trial species for the 24 questions ranged from 0-23 with an average score of 12 (Table 2). The range of scores for plants known to be invasive in NSW ranged from 14-23, with an average score of 19.

To evaluate the test results, we separated the trial species into two categories: invasive anywhere in the world or not known to be invasive anywhere in the world. Overall, the tool achieved 100% accuracy in accepting non-invasive species and rejecting invasive species, excluding the four species requiring further evaluation: Cyperus esculentus, Prunus persica, Cosmos sulphureus and Viburnum tinusise (Figure 2; Table 2).

Figure 2. Frequency histogram of the scores of the 58 species evaluated in the trial of the Plant Sure Ornamental Plant Decision Support Tool divided into categories of invasive or non- invasive to assess accuracy of the tool. Invasive species are those that have been reported as being invasive anywhere globally. Non-invasive represents species not reported as invasive anywhere globally.

12

Testing the accuracy of plants not known to be invasive in NSW, particularly those that are naturalised but not invasive, was less straightforward. Of the 18 naturalised species, half were accepted and half were rejected (including one species to evaluate further). Of the 24 non-naturalised species, 21 were accepted, 2 were rejected and 1 needed further evaluation (Figure 3). Most of the naturalised and non-naturalised species were rejected due to their invasiveness in places outside Australia with similar climates to NSW. The scoring was low for almost all of the accepted naturalised species (8 or less), due to no current evidence of the species being invasive elsewhere, indicating that the risk of these species becoming invasive in NSW is low. However, species that are naturalised may become invasive in the future, so caution is necessary. To alleviate any concerns that an accepted naturalised species may become invasive in the future, a clause could be inserted that an accepted naturalised species must be reviewed by more than one assessor and decisions are made on a case by case basis.

Figure 3. The scores of the 58 species evaluated in the trial of the Plant Sure Ornamental Plant Decision Support Tool, categorised in NSW as invasive, naturalised and non-naturalised

13

The pre-screening tool performed well as a rapid assessment tool. Overall, the total score and pre-screening score were highly correlated (r = 0.97; Figure 4). The pre- screen rejected 15 out of the 16 invasive species in the first 6 questions (Table 3). The known invasive species that was not rejected in the pre-screen (Prunus persica) was categorised as evaluate further when undergoing the full screen evaluation.

Table 3. Comparison of the pre-screen and main screen results.

Pre-screen Main screen

Accept Reject Accept Evaluate Reject further

Invasive 1 15 2 14 Naturalised 10 8 10 1 7 Non- 22 2 21 1 2 naturalised

Total 33 25 31 4 23

Figure 4. Scatter plot showing the correlation between the pre-screen score and the main screen score for the 58 plants assessed in the trial of the Plant Sure Ornamental Plant Decision Support Tool.

14

There was a high average response rate for all questions (90%), ranging from 49 – 100% (Table 4). The question with the lowest response rate (49%) was: “Is there significant germination (>25%) of seeds the next growing season, with no requirement of an infrequent environmental condition for seeds to germinate (i.e. fire) or long dormancy period?” Assessments are invalid if less than 80% of the questions are unanswered.

Time taken to complete the plant risk assessments ranged from 10 minutes to 90 minutes for an experienced weed risk assessor and from 2 to 4 hours for someone with less experience in assessing weed risk (overall average time of 115 minutes). The time taken also depended on how easily information could be found for the species and was reduced when a compilation of relevant web sites was made available for the assessment process. This list comprises Commonwealth and State noxious and prohibited plants and an assortment of websites known to list invasive and naturalised species in Australia and internationally (Table 1 and Appendix 1). All of the trial users had at least a medium level of plant knowledge; therefore, a person with less botanical knowledge should expect an assessment to take longer than 1-2 hours.

Table 4. Response rates of the 24 questions included in the Plant Sure Ornamental Plant Decision Support Tool.

Question Respo nse rate 1) Is this species a cultivar or variety? If yes, assess this species using the cultivar 98% assessment tool. If not, proceed with questions below. 2) Is this species listed on any Commonwealth or state noxious or prohibited plant lists? If 98% yes, stop assessment. 3) Is the species noted as being naturalised in Australia and/or the world in a similar 98% climate? 4) Is the species noted as being invasive in Australia and/or the world? 98% 5) Is the species noted as being invasive in Australia and/or the world in a similar climate? 98% 6) Does the species displace native plants and/or dominate (overtop or smother) the plant 98% community in areas where it has established? 7) Does the species reproduce and spread vegetatively? 100% 8) Does the species produce copious viable seeds each year (> 1000)? 89% 9) Has the species become naturalised where it is not native? 98% 10) Are other species of the same genus (or very closely related genera) invasive in a similar 95% climate? 11) Is the species native range in a climate matching the region of concern? 97% 12) Is the species noted as promoting fire and/or changing fire regimes? 97% 13) Is the species a health risk to humans or animals/fish? Has the species been noted as 97% impacting grazing systems? 14) Does the species produce impenetrable thickets, blocking or slowing movement of 95% animals, livestock, or humans? 15) If naturally detached fragments from this plant are capable of producing new plants, is 98% this a common method of reproduction for the plant? 16) Does the species (or cultivar or variety) commonly produce viable seed? 89%

15

17) Is there significant germination (>25%) of seeds the next growing season, with no 49% requirement of an infrequent environmental condition for seeds to germinate (i.e. fire) or long dormancy period? 18) Does this plant produce viable seed within the first three years (for an herbaceous 69% species) to five years (for a woody species) after germination? 19) Does this plant continuously produce seed for >3 months each year or does seed 72% production occur more than once a year? 20) Are the plant’s propagules frequently dispersed long distance (>100 m) by mammals or 95% birds or via domestic animals? 21) Are the plant’s propagules frequently dispersed long distance (>100 m) by wind or water? 95% 22) Are the plant’s propagules frequently dispersed via contaminated seed (agriculture or 95% wildflower packets), equipment, vehicles, boats or clothing/shoes? 23) Is the distribution of this species projected to increase under future climate change in 69% NSW? 24) Is there evidence of a species in this genus hybridising with a species that is invasive in 67% NSW?

During the trials, we reduced repetition between questions, culminating in 24 final questions. Where questions appeared to be repetitive we measured the correlation of the answers to these questions. Some of the remaining questions may still appear to be repetitive but the correlation was low and these questions have therefore been retained. Some minor modifications were made to the wording of those questions to better separate their meaning. For example the correlation between Q.7 “Does the species produce copious viable seeds each year (> 1000)? and Q.15 “Does the species (or cultivar or variety) commonly produce viable seed?” was 18%. The correlation between Q.8 “Does the species reproduce and spread vegetatively?” and Q.16 “If naturally detached fragments from this plant are capable of producing new plants, is this a common method of reproduction for the plant?” was 80%.

The three cultivars that were tested using both the main and cultivar screening tools received the same overall result. Unless further cultivar testing identifies a higher degree of sensitivity in accurately evaluating cultivars, the main screening tool could perform just as well in cultivar assessments.

Each question in the OPDST asks for a confidence rating of the answer (high, medium or low confidence). After completing an evaluation, an overall confidence level could be given for the rating based on the individual confidence levels for answering each question, along with a justification for the level of confidence. For example, a species that lacks information on the vegetative propagation potential should be given a low confidence rating with an explanation of “a review of the assessment should be completed in 3 years or if new information becomes available on the vegetation propagation success of the species; whichever comes first.” Additional fields to record reference sources and any additional comments are provided and these will be helpful in assessing those species that need further evaluation and at the time of reassessment.

16

Using confidence levels in conjunction with the overall evaluation scores gives the end user, particularly non-experts, the ability to quickly interpret the results. As an example, if a species receives a score of 10 out of a possible 24 points, it is deemed as non- invasive and therefore low risk. If all of the answers were allocated “high confidence”, the species position in the matrix would be in the top right hand corner (see Table 5) and this species would be given a green light for sale. Species that receive a low confidence rating should be targeted for frequent reassessment. It is important to note that at least 16 questions (80%) must be answered for the assessment to be considered "valid".

Table 5. Example of a risk matrix to be completed after the evaluation process

Low Medium High confidence confidence confidence

Low risk x

Medium risk

High risk

17

Recommendations

As with any predictive or assessment tool, there are a certain amount of limitations and assumptions that are made. To minimise any shortcomings of the Plant Sure OPDST, the following recommendations are suggested;

I. Naturalised species in NSW that are accepted by the OPDST are reviewed by more than one assessor on a case by case basis; II. An expert group of OPDST users should determine the duration of assessment before requiring re-validation. For example, assessments for species categorised as: ‘non- invasive’ could be re-assessed every 3-5 years; ’evaluate further’ every 2-3 years; and ‘invasive’ every 5 years; III. The reliability and accuracy of the OPDST depends on the quality of the data. Relevant data can be difficult to find and it is therefore preferable that an expert panel, who have access to research libraries and databases, and have reliable internet access, complete and verify the evaluations; IV. Assessments, completed by the expert panel, should be made available to all end users; V. All weed list databases (provided in the OPDST spreadsheet, Appendix 1) should be regularly updated and maintained; VI. The species listed in the noxious and prohibited plant lists are consolidated into one source (see Web sites worksheet, Appendix 1). Currently, the user accesses 10 websites. This consolidation will reduce the time taken and the possibility of error; VII. Where a species receives a pre-screen total of 7 out of a possible 8 points, the evaluation should be reviewed by a second person; VIII. The pre-screen questions are designed as a time saving process for those species that are rejected. The assessor does not need to evaluate any further if a total of ≥8 is reached but it is imperative that the full assessment is completed where the total is below this score; IX. To comply with the ‘tens rule’ and to avoid sampling bias, further assessments should be undertaken using the main tool, with priority given to non-invasive species; X. Only three cultivars were assessed in the trials and they received the same output under both the OPDST and the separate cultivar tool. Further trials should be undertaken to more fully compare the accuracy of the two tools; XI. The OPDST should NOT be used to evaluate species endemic to the region, or aquatic species; XII. Additional work is needed to refine target region (NSW) into finer-scale regions; XIII. Consideration should be given to building a climate matching tool particularly for NSW / Australia as per the Plant Right Evaluation Tool in California that divides NSW into bioclimatic regions; XIV. The evaluation (< 13 Accept; 13 – 15 Evaluate further; > 15 Reject) is invalid if ≤ 15 questions were answered because of insufficient information to make a robust assessment;

18

XV. If and when the Weed Futures tool (http://www.weedfutures.net/) has been updated to include all introduced species, question 23 “Is the distribution of this species projected to increase under future climate change in the target region?” should be included in the score; XVI. The confidence level, Justification, References(s) and Additional notes columns in the OPDST should be completed because this will assist in future reassessments; XVII. Whilst the OPDST is relatively easy to use and straightforward, a series of workshops for end-users would facilitate its uptake; XVIII. Retesting for accuracy of the tool should be undertaken if substantial changes to the tool are requested by stakeholders; XIX. Consensus is reached amongst stakeholders on the definition of ‘invasive’ and whether further description of the invasive process is desirable e.g. Appendix 4, supplied by Steve Taylor, Invasive Plants Coordinator, ACT Parks and Conservation Service, and Stephen Johnson, Weed Ecologist, (Weeds Incursions and Response), Weed Research Unit, Department of Primary Industries (NSW); and XX. Further to XIX, feedback from the stakeholder workshop held on 23/2/18 indicated that ‘the impacts on human health’ should be considered more thoroughly and separately rather than its current inclusion as part of the definition of an invasive species. Consensus of the treatment of this definition and the location of its inclusion in the glossary should be sought from stakeholders.

19

References

Brown, L., Barry, S., Cunningham, D. & Bomford, M. (2006) Current practice in applying CLIMATE for weed risk assessment in Australia. Proceedings of the 15th Australian Weeds Conference, Adelaide, South Australia, 2006. 703-706. Conser, C., Seebacher, L., Fujino, D. W., Reichard, S. & Ditomaso, J. M. (2015). The Development of a Plant Risk Evaluation (PRE) Tool for Assessing the Invasive Potential of Ornamental Plants. PLoS One 10, e0121053 https://doi.org/10.1371/journal.pone.0121053 Coutts-Smith, A., Downey, P., (2006) Impact of weeds on threatened biodiversity in New South Wales, Technical Series No. 11. Cooperative Research Centre for Australian Weed Management, Adelaide. Cresswell, I.D. & Murphy, H.T. (2017). Australia state of the environment 2016: biodiversity, independent report to the Australian Government Minister for the Environment and Energy, Australian Government Department of the Environment and Energy, Canberra. Dodd, A. J., Burgman, M. A., McCarthy, M, A, Ainsworth, N. (2015). The changing patterns of plant naturalization in Australia. Diversity and Distributions 21, 1038- 1050. Eyre, D., Baker, R. & Brunel, S. (2013). Matching methods to produce maps for pest risk analysis to resources. NeoBiota, 18, 25. Eyre, D., Baker, R. H., Brunel, S., Dupin, M., Jarošík, V., Kriticos, D. J., Makowski, D., Pergl, J., Reynaud, P. & Robinet, C. (2012). Rating and mapping the suitability of the climate for pest risk analysis. EPPO Bulletin, 42, 48-55. Gallagher, R.V. & Leishman, M.R. (2015) Invasive plants and invaded ecosystems in Australia: implications for biodiversity. Austral ark: the state of wildlife in Australia and New Zealand (ed. by A. Stow, N. Maclean and G.I. Holwell), pp. 105–133. Cambridge University Press, Cambridge. Hulme, P. E. (2012). Weed risk assessment : a way forward or a waste of time. Journal of Applied Ecology 49, 10-19. Invasive Species Council (ISC), 2009, https://invasives.org.au/, accessed 9/12/2017 Jordan, M.J., Moore, G., Weldy, T.W. (2012). New York State Ranking System for Evaluating Non-Native Plant Species for Invasiveness. Available at: http://www.nyis.info/pdf/New_York_State_Invasive_Plant_Ranking_System_Rev_2 012.pdf (Accessed September 2017) Lieurance, D., Flory, S.F., Gordon, D.R. (2013). The UF/IFAS Assessment of Nonnative Plants in Florida’s Natural Areas: History, Purpose, and Use. Document SS-AGR- 371. Agronomy Department, UF/IFAS Extension, Available at: https://assessment.ifas.ufl.edu/. (Accessed September 2017). Maher, T., Graham, V., Hancock, N., Adams, V. & Leishman, M. (2017). Review of National & International Weed Risk Assessment Tools for the ‘Plant Sure Environmentally Safe Ornamental Plant Scheme’ project. Macquarie University

20

Metcalfe, D.J. & Bui, E.N. (2017). Australia state of the environment 2016: land, independent report to the Australian Government Minister for the Environment and Energy, Australian Government Department of the Environment and Energy, Canberra, doi:10.4226/94/58b6585f94911. Pheloung, P. C., Williams, P. A. & Halloy, S. R. (1999). A weed risk assessment model for use as a biosecurity tool evaluating plant introductions. Journal of Environmental Management, 57, 239-51. Scott, J.K., Webber, B.L., Murphy, H., Ota, N.,Kriticos, D.J., Loechel, B. (2014) AdaptNRM Weeds and climate change: supporting weed management adaptation. Available at:www.AdaptNRM.org. Sinden, J., Jones, R., Hester, S., Odom, D., Kalisch, C., James, R., Cacho, O., (2004) The economic impact of weeds in Australia, Technical Series No. 8. Cooperative Research Centre for Weed Management Systems, Adelaide. Virtue, J.G., (2008) SA Weed Risk Management Guide. Department of Water Land & Biodiversity Conservation, South Australia. Weiss, J.R., McLaren, D.A., (2002) Victoria’s pest prioritisation process, in: Spafford Jacob, H., Dodd, J., Moore, J.H. (Eds.), 13th Australian Weeds Conference proceedings: weeds 'threats now and forever'. Plant Protection Society of Western Australia, Perth, pp. p. 509-512.

21

Appendix 1

OPDST questions and scoring mechanism; instructions; cultivar tool; example; web sites; glossary Provided as an attachment.

22

Appendix 2

Climatch: Instructions on how to use Climatch

Q.3) Is the species (or cultivar or variety) noted as being naturalized in Australia or the world in a similar climate?

Answer Yes (2 points) if the species is naturalised in South-eastern Australia / NSW.

If now, use Climatch to address whether the species is naturalised in another part of the world in a similar climate (as per below instructions).

1.) Do a web search for information on which countries and/or states the target species is known to be naturalised in. (GBIF: https://www.gbif.org/ and Invasive Plant Atlas are good sources to find species occurrences for this question). 2.) Go to the online Climatch tool: http://data.daff.gov.au:8080/Climatch/climatch.jsp 3.) In Climatch click on the "Source Region" tab and scroll down to "Data set" and click on the drop down menu and select "Worldclim Sample".

4.) From here, use the zoom and select tools to highlight all of the blue dots in the “Source Map” within the borders of the countries and/or states that your target species is known to be naturalised in. Note: the GBIF database doesn’t distinguish between records inside the native range or outside the native range. Use online resources to help identify where the naturalised (non-native) range is and highlight only those regions in Climatch.

23

5.) Scroll down and click the hammer symbol and select the match settings for the following climate variables.

6.) Select the “Target Region” tab and scroll down and select “Worldclim Sample” from the drop down menu.

7.) Use the zoom and select tools to highlight all of the blue dots within NSW.

24

8.) Click on to see which areas of NSW have a similar climate to the original area that you selected for the target species. This should result in the following map with a grading of increasing climate suitability from 1-10.

9.) Once you have obtained this climate similarity map, please click on the save symbol at the bottom of the map and save map as a .png image for future reference 10.) If any region of NSW recieves a climate suitability rating of 5 or above answer Yes and give 2 points.

Q.5) Is the species (or cultivar or variety) noted as being invasive in the NSW or world in a similar climate? (similar method to question above) Use Climatch to address whether the species is invasive in another part of the world in a similar climate.

1) To answer this question, repeat the same steps as detailed for in Question 3.) except in the "Source Region" tab in Climatch, select all blue dots in the states and or countries where the target species is known to be invasive. Note: If the species is only invasive in part of a state or country, we still select

25

all the blue dots in that state and/or country because it may be that the species hasn’t been established for long and may expand its invasive range within these areas with time. Q.10) Are other species (or cultivars or varieties) of the same genus (or very closely related genera) invasive in a similar climate to NSW?

1. Do a web search to check to see if there are any other important invasive cultivars/varieties of this species or other invasive species in the same genus. If not, allocate 0 points to this question. 2. If yes, use the Climatch tool: http://data.daff.gov.au:8080/Climatch/climatch.jsp to select all the blue dots corresponding to countries and/or states in which this related species is known to be invasive and check to see how this matches with the climate in NSW using the same method detailed for the previous questions. Note:  If the species is clearly similar in and name to another important invasive species, that is known to be invasive in a similar climate to NSW, then you can answer the first part of the question yes. However, some species have changed names or synonyms that are in the same genus or very closely related to an invasive plant. If there are no close relatives, then the answer should be no.  The Global Compendium of Weeds is a good reference to use for this question. If no other species in the genus is listed in the Compendium, then it is fairly safe to assume that none are invasive. The Compendium must be used with the codes and references specified at the beginning of the book. Many species are cited with an “N” which means naturalized only. Please find the attached PDF version of the book (again, you need to be familiar with the codes it uses – see page 3 of this PDF for a key). The answer will be either yes or no, and should never go unanswered.

Q.11) Is the species native range in a climate matching the region of concern?

1) To answer this question, repeat the same steps as detailed for in Question 3.) except in the "Source Region" tab in Climatch, select all blue dots in the states and or countries where the target species is known to be native. The answer will be either yes or no, and should never go unanswered.

26

Appendix 3

New York State 6 NUYCRR Part 575 Invasive Species Cultivar Invasiveness Assessment Protocol Provided as a pdf. Attachment

27

Appendix 4

Supplied by Steve Taylor, Invasive Plants Coordinator, ACT Parks and Conservation Service, and Stephen Johnson, Weed Ecologist, (Weeds Incursions and Response), Weed Research Unit, Department of Primary Industries (NSW)

28