A weed risk assessment model for use as a biosecurity tool evaluating plant introductions
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Journal of Environmental Management (1999) 57, 239–251 Article No. jema.1999.0297, available online at http://www.idealibrary.com on A weed risk assessment model for use as a biosecurity tool evaluating plant introductions P. C. Pheloung†‡ , P. A. Williams§* , S. R. Halloy¶ New plant taxa from around the world continue to be imported into Australia and New Zealand. Many of these taxa have the potential to become agricultural or environmental weeds and this risk needs to be assessed before allowing their entry. A weed risk assessment system is described that uses information on a taxon’s current weed status in other parts of the world, climate and environmental preferences, and biological attributes. The system is designed to be operated by quarantine personnel via a user-friendly computer interface. The model was tested against experts’ scores for weediness for 370 taxa present in Australia, representing both weeds and useful taxa from agriculture, the environment, and other sectors. The model was judged on its ability to correctly ‘reject’ weeds, ‘accept’ non-weeds, and generate a low proportion of taxa which could not be decisively categorised, termed ‘evaluate’. More than 70% of the taxa were rejected or accepted. All taxa classified as serious weeds, and most minor weeds, were rejected or required further evaluation, while only 7% of non-weeds were rejected. The model was modified to New Zealand conditions and evaluated against the opinions of several groups of experts and against economic measures. The model produced a weediness score very similar to the mean of the experts scores. The latter were highly variable: agriculturalists tended to accept known weeds, conservationists tended to reject most adventive taxa, and only botanists produced scores similar to the model. The model scores also tended to be independent of economic value as measured in this study. The model could be adapted for use as a screening tool in any region of the world. 1999 Academic Press Keywords: Australia, New Zealand, weed risk assessment, plant introductions, biosecurity, quarantine, computer model. Introduction potential (reviewed by Mack, 1996), but bor- der authorities urgently need an objective, Ł Corresponding author credible, and publicly acceptable risk assess- Weeds have major impacts on economies and ment system to predict the weediness, or † Agriculture Western natural environments world wide, including invasive potential, of the thousands of poten- Australia, 3 Baron-Hay Crt, South Perth WA 6151, Australia (Groves, 1986) and New Zealand tial new entries. Australia (Williams and Timmins, 1990). Many of these There have been several symposia (e.g. weeds have been purposely introduced as Drake et al., 1989) and a growing research ‡ Present address: Australian Quarantine potential new crops, ornamentals, or as novel- literature on invasive plant taxa, but until Inspection Service, GPO ties (Esler, 1988; Lonsdale, 1994). To counter recently there was considerable pessimism Box 858, Canberra, the threat to agriculture or the environment that potentially invasive species could be ACT 2601, Australia from new plant taxa, regulatory authori- identified (Crawley, 1987). However, progress § Landcare Research, ties have a statutory responsibility to ensure has recently been made in identifying the Private Bag 6, Nelson, that all plant taxa proposed to be imported, key characteristics of potential weeds in New Zealand which are not already prohibited and are not Australia (Noble, 1988; Scott and Panetta, ¶ Crop and Food, Private already established, be evaluated for their 1993), New Zealand (Esler et al., 1993), Bag, Mosgiel, New potential to damage the productive capac- South Africa (Richardson et al., 1990), Zealand ity or environment of these countries. There Britain (Williamson and Fitter, 1996), and Received 27 August 1998; are many approaches to predicting weed north America (Mack 1996; Reichard and accepted 23 April 1999 0301–4797/99/120239C13 $30.00/0 1999 Academic Press
240 P. C. Pheloung et al. Hamilton, 1997), or within closely related authority that satisfies many of these crite- groups of taxa including herbs (Forcella et al., ria: the Weed Risk Assessment model (WRA). 1986; Williamson, 1993) or woody seed plants We then report on a test of the applicabil- (Rejmanek and Richardson, 1996). Predic- ity of a slightly altered version of the model tions from a non-experimental approach are that incorporates environmental differences most accurate if they are based on informa- between Australia and New Zealand: the tion that includes the biology of the invad- New Zealand Weed Risk Assessment Model ing species, the bioclimatic features of the (New Zealand WRA). In testing the New recipient and source regions, and the evo- Zealand WRA we examine the sources of lutionary history of both (e.g. Richardson expert bias that are inherent in the con- et al., 1990). A synthesis of this informa- cept of weediness (Perrins et al., 1992) and tion can lead to general conceptual models of demonstrate that the model minimises these plant invasiveness (Rejmanek, 1995) which difficulties. in turn can be applied to screening procedures The model could be used to investigate such as decision trees to categorise poten- the nature of weediness, but here we simply tial plant imports (Panetta, 1993; Tucker and describe how it works and the steps we took Richardson, 1995; Reichard, 1996; Reichard to ensure it would function as a publicly acceptable screening tool. and Hamilton, 1997). Regulatory authorities responsible for biosecurity need a tool that synthesizes much of this disparate and some- Methods and results times theoretical information. One approach is a computer based spreadsheet model pro- Producing the WRA ducing a score for weediness, which can then be converted to an entry recommendation for The basis of the WRA is the answers to a specified taxon. 49 questions (Appendix 1) based on the main An acceptable biosecurity assessment sys- attributes and impacts of weeds. These are tem should satisfy a number of requirements combined into a scoring system which in (Hazard, 1988; Panetta, 1993). It should be the absence of any evidence to the contrary, calibrated and validated against a large num- gives an equal weight to nearly all questions ber of taxa already present in the recipient (Appendix 2). These cover a range of weedy country and representing the full spectrum of attributes in order to screen for taxa that are taxa likely to be encountered as imports into likely to become weeds of the environment that country. It must effectively discriminate and/or agriculture. The questions are divided into three sections producing identifiable between weeds and non-weeds, such that the scores that contribute to the total score majority of weeds are not accepted, non- (Appendix 2). weeds are not rejected, and the proportion Biogeography (section A) encompasses of taxa requiring further evaluation is kept the documented distribution, climate to a minimum. As international trade agree- preferences, history of cultivation, and ments require that prohibited taxa should weediness of a plant taxon elsewhere in the fit the definition of a quarantine pest before world, i.e. apart from the proposed recipient they can be excluded by quarantine regula- country. Weediness elsewhere is a good tions (Anon., 1994, 1995; Walton and Parnell, predictor of a taxon becoming a weed in new 1996), the system must be based on explicit areas with similar environmental conditions assumptions and scientific principles so that (Forcella and Wood, 1984; Panetta and the country cannot be accused of applying Mitchell, 1991a,b). The question concerning unjustified non-tariff trade barriers. Ideally the history of cultivation recognises the the system should be capable of identifying important human component of propagule which land use system the taxon is likely to pressure (Richardson et al., 1994; Williamson invade, to assist in an economic evaluation and Fitter, 1996), but such data are obviously of its potential impacts. Finally, the sys- never available for the proposed new tem must be cost effective to the prospective country. The global distribution and climate importer and the border control authority. preferences, where these are available, are Here, we first describe a model designed used to predict a potential distribution in the specifically for the Australian quarantine recipient country.
A weed risk assessment model 241 Undesirable attributes (section B) are char- A benchmark against which to compare the acteristics such as toxic fruits and unpalat- model is problematic because there are no ibility, or invasive behaviour, such as a absolute values for weediness of individual climbing or smothering growth habit, or the taxa (Perrins et al., 1992). However, those ability to survive in dense shade. familiar with a taxon in a country can offer a Biology/ecology (section C) are the attri- meaningful opinion on the actual or potential butes that enable a taxon to reproduce, weediness of the taxon in that country, spread, and persist (Noble, 1988), such as against which to compare the score from the whether the plant is wind dispersed or animal model. (The sources of bias in this procedure dispersed, and whether the seeds would are examined in evaluating the New Zealand survive passage through an animal’s gut. WRA.) A further 12 scientists therefore Availability of information is often very defined the weed status or usefulness of taxa limited for new species which can restrain they were familiar with. Taxa were given a the utility of screening systems. To ensure rank of 0, 1 or 2 and this was used to classify that at least some questions were answered them as non-weeds, minor weeds, and serious for each section, the WRA system requires weeds, respectively. The mean score was used the answers to two questions in section A, to assess the performance of the WRA model; two in section B, and six in section C before it again, potential bias is examined in testing will give an evaluation and recommendation. the New Zealand WRA. The recommendation can be compared with the number of questions answered as an indication of its reliability, which obviously Results of the WRA improves as more questions are answered. Answers to the questions provide a poten- The majority of the 370 taxa (81%) are tial total score ranging from 14 (benign perceived as weeds in some context, but taxa) to 29 (maximum weediness) for each less than half are considered serious weeds. taxon. The total score is partitioned between Useful taxa (63%) including both weeds answers to questions considered to relate pri- (45%) and non-weeds (18%), and non-useful marily to agriculture, to the environment, or taxa (37%) were included in the sam- common to both (Appendix 1). The total scores ple. Weeds from all sectors—agriculture, are converted to one of the three possible rec- environment, horticulture, garden and ser- ommendations by two critical score settings. vice areas—were well represented (data not The lower critical score, 0, separates accept- shown). able taxa from those requiring evaluation, The cumulative frequency distributions of and the higher critical score, 6, separates WRA scores, for each of the survey classifica- taxa requiring evaluation from those that tions (Figure 1), were used to investigate the should be rejected. Evaluation could mean effect of different pairs of critical scores on either obtaining more data and re-running the distribution of WRA recommendations. the model, or undertaking further investiga- The range of scores for non-weeds overlaps tions such as field trials (Mack, 1996). the range for serious weeds, so it is impos- To provide data for defining these score sible to define any set of critical scores that settings, six Australian scientists in a range reject all serious weeds while accepting all of the fields ran the model to assess the non-weeds. However, it is possible to ensure weed potential of taxa they were familiar that none of the serious weeds are accepted by with, ranging from non-weedy beneficial taxa setting the accept score at 0 or less. Similarly, to serious weeds. In addition, all taxa that less than 10% of non-weeds will be rejected if have a noxious status in Australia were the reject score is greater than 6. This would assessed using the information provided by mean that 29% of the taxa assessed in this Parsons and Cuthbertson (1992). Assessors study would fall between these extremes and treated each taxon as if it had not yet require evaluation. Lowering the minimum arrived in Australia, such that serious weeds, reject score to 6 would reduce this proportion for example, were assessed purely on their to 22% but increase the proportion of rejected weed status outside Australia. The taxa were non-weeds to 15%, which is less desirable, classified as agricultural or environmental since some of these are regarded as useful weeds. (Figure 1). Consequently, the critical scores,
242 P. C. Pheloung et al. Proportion of species scoring the amount shown or less (%) 100 80 60 Accept Reject 40 20 0 –15 –10 –5 0 5 10 15 20 25 30 Weed risk assessment score Figure 1. Cumulative frequency of taxa receiving a particular WRA score or less for each survey classification. Survey categories: ( M ), serious weeds (139); ( ž ), minor weeds (147); ( ), non-weeds (84); ( ), all plants (370). 0 and 6, were used to convert the WRA scores Table 2. Correlations of the agriculture and into the recommendations–‘accept’, ‘evaluate’ environment components of the WRA score and the total WRA score, with the experts and ‘reject’. These settings recommended all classification of taxa as weeds of agriculture serious weeds and most minor weeds (84%) or environment. All P
A weed risk assessment model 243 ‘a range of soil conditions’ because infertile The WRA database contains replicate soils are not as extensive in New Zealand. As scores from the six independent Australian anticipated, several questions that remained experts. These were used to estimate the vari- the same needed to be answered differently ance of the model scores, on the assumption for Australia and New Zealand; e.g. in that a similar degree of variability would be Australia there are many parrots which found in the scores from experts using the destroy seeds, and so certain new plants New Zealand WRA model. are more likely to have natural enemies in Next, 13 New Zealand experts, 12 of whom Australia than in New Zealand. None of were not involved with the model, classified the modified questions required alterations many of the same 291 taxa into the categories to the scoring system, so that the outcome of non-weed, minor weed, and major weed. scores were directly comparable. The outcome The categories were not defined, and neither of these changes was the New Zealand were the taxa classified as environmental or WRA model. agricultural weeds as in Australia. However, respondents were asked not to consider economic value, i.e. we were interested only in their opinion as to whether taxa were weeds Testing the New Zealand WRA in any land-use system. This classification was transformed into the numbers 0, 0Ð5 and We assessed the ability of the model to iden- 1Ð0, respectively. The experts were classified tify weeds of a range of land use systems when by one of us (P.A.W.) as botanists (seven), used in a country for which it was not origi- agriculturalists (three), and conservationists nally designed, in this instance New Zealand. (three). This classification was unknown to We also tested the New Zealand WRA—in them, to avoid any conscious bias. This a manner that was equally relevant to allowed an evaluation of speciality bias with WRA—by asking whether it was better than respect to classification, and how it affected an expert’s opinion? ‘Better’ we define as variation. Each expert reviewed between efficacy, ease of use, and cost of use but only twenty-seven and 291 taxa, with most doing efficacy was tested, i.e. the capability to score between 140 and 250 taxa. Two hundred and correctly with respect to the benchmark, and seventy-four taxa were classified by more with a low variation when used by several than one expert. individuals. Finally, 195 taxa for which data were avail- We tested whether the New Zealand WRA able were scored for relative economic value, was more biased by economic or environ- based on the average of three evaluations. mental considerations than the collective (1) The area of the taxon cultivated from opinion, as represented by an average score, 1842 to 1991 was obtained from New Zealand of a group of experts representing a wide statistics as processed by Halloy (1994). The range of interests. This potentially large ele- logarithm of this area was standardised to a ment of subjectivity was tested with a series range of 0 to 1. (2) An estimate of an eco- of cross-checks on variation and bias using nomic value of any type was scored as 0 to 5 three data sets referred to the same taxa. by one of us (S.R.H.) then standardised 0 to 1. Firstly, the New Zealand WRA model was (3) Taxa were scored according to their pres- run on 198 taxa used in testing the WRA ence or absence in the New Zealand Nursery model that are also present in New Zealand Register 1995/1996, on the assumption that (the majority of which are adventive), plus a listing represents value as an ornamental any additional noxious plants in New Zealand but not as a major economic plant. Scores (Esler et al., 1993). Responses to the model were 1 if the taxon was present, 0Ð5 if only questions for the resulting 291 taxa were the genus was present but with probability of made by one of us (P.A.W) as if they were not the taxon being present, 0 if the genus was yet in New Zealand, using the international absent. This value was standardised 0 to 0Ð2. literature, floras, and databases. The scores The data were divided into five classes, and of 14 to 29 for each taxon were standardised the taxon score outcomes for New Zealand from 0 to 1 for comparison with the other data WRA and the experts were compared within sets, using the thresholds: accept
244 P. C. Pheloung et al. WRA within each economic class is expressed ‘evaluate’ category of the three expert groups, as percentages. as against 21% for the model, shows that Lastly, the total scores for the WRA and all expert groups have a tendency to either New Zealand WRA models derived in the two accept or reject a larger proportion of taxa countries were compared for the 198 taxa than the model (Table 3). common to both. There are differences in recommendations between the New Zealand WRA model and the average experts’ evaluation (Figure 2). Results of testing the New Zealand Sixty-four percent of taxa did not change, WRA and overall 93% of taxa either rejected by the model or requiring evaluation were The combined scores and their variability also not accepted by the experts. Only 4% show a strong similarity between the scores of taxa that were accepted by the model from the New Zealand WRA model (0Ð538) were rejected by the expert averages. Two and the mean scores for the 13 experts differences between the New Zealand WRA (0Ð545), and for the deviations, and the dis- outcomes and the expert outcomes (Figure 2) tributions (Table 3). When converted to out- require examination. comes the results are again very similar. That the higher proportion of taxa rejected by both (1) A greater permissiveness in the model the New Zealand WRA model and the experts as compared with the experts could allow score (mean 62Ð0%) is higher than might be introduction of weeds if the model alone were expected from a random selection of taxa used as a screening tool. Eight percent of partly reflects the addition of all the noxious taxa were accepted by the model but classed weeds of New Zealand to the database. as ‘reject’ or ‘evaluate’ by the expert averages. Replicate answers to the New Zealand Taxa accepted by the model but rejected by WRA model are less variable (CV 8Ð2%) experts include: Chasmanthe floribunda (Sal- than the mean expert scores (CV 58Ð7%) isb.) N. E. Brown, Echium candicans L., Era- (Table 3). The high variation in taxon scores grostis tef Trott and Scabiosa atropurpurea is explained by the variability among the L. Taxa rated as ‘evaluate’ include Betula individual respondents, even within expert pendula Roth, Festuca rubra L., Gliricidia groups. Agriculturalists gave the lowest sepium (Jacq.) Walp and Lathyrus ochrus L. scores for weediness (0Ð438), conservationists These taxa have several features in common. the highest (0Ð673). Conservationists’ scores They received a low number of responses were also less variable than those of the (two or three), their expert score variation other groups (19Ð5% cf. 31Ð3% and 33Ð8%). was high (data not presented), and most had The outcome is that conservationists tend to model scores on or near the thresholds. It reject most taxa while agronomists accept a is significant that no taxon with an average higher proportion. The 5–6% of taxa in the expert score of major weed was accepted by Table 3. Comparisons of scores from NZWRA and the expert groups: (a) the average of all taxa scored and the variation; (b) the average level of variability of scores for individual taxa from NZWRA and the experts for taxa with two or more scores; (c) the outcomes for all taxa expressed as a rounded percentage NZ Model Mean Botanists Agriculturalists Conservationists expert (a) All taxa scores N 291 291 288 251 291 Mean 0Ð538 0Ð545 0Ð520 0Ð438 0Ð673 SD taxa score 0Ð204 0Ð238 0Ð307 0Ð311 0Ð258 (b) Individual taxa N 291 291 208 181 224 Mean SD of score 0Ð041 0Ð231 0Ð115 0Ð099 0Ð087 CV % of score 8Ð2 58Ð7 31Ð3 33Ð8 19Ð5 (c) All taxa outcomes Accept 16 17 25 35 7 Evaluate 21 20 5 6 4 Reject 63 63 70 59 89
A weed risk assessment model 245 80 70 Proportion of model outcomes which differed from experts outcomes 60 50 40 30 20 10 0 Expert: No change Evaluate Reject Accept Reject Accept Evaluate Model: Accept Evaluate Reject Figure 2. A comparison of the outcomes from the NZWRA model vs. the outcomes from the expert groups, expressed as a percentage of the total outcomes for each group. , agriculturalists (251); , botanists (288); , conservationists (291); , all (291). the New Zealand WRA model, nor any taxon Scoring by the model and by the experts rated as a noxious plant in New Zealand. varied according to the economic value of (2) A greater severity of assessment by the the taxon (Figure 3). The mean expert score model than by the experts could result in showed very little bias, with only a slight benign taxa being rejected. The model was tendency to accept more taxa (16Ð7%) of more cautious than the experts in 19% of intermediate economic value (
246 P. C. Pheloung et al. 80 Difference in accepted vs. rejected (%) 60 40 20 0 –20 –40 –60 0
A weed risk assessment model 247 availability of experts may be limiting, but Zealand context Agrostis stolonifera L., Lupi- any smaller number of experts—especially if nus polyphyllus and Pinus contorta. Where they were from one interest group—would a taxon may have significant economic ben- be likely to give a biased answer with efits, a further evaluation of its weediness respect to a particular sector (Perrins et al., potential may include experimental stud- 1992). ies (Williamson, 1993; Scott and Panetta, The bias displayed by different groups of 1993). Economic value should be scored in experts differs between the two countries for a transparently separate exercise and bal- which we have information. Conservation- anced against weediness in appropriate risk ists in Britain, for example, are less inclined assessment evaluations (Walton and Parnell, than the average to consider a taxon to be 1996). a weed (Perrins et al., 1992; Williamson, We conclude that the weed risk assess- 1993), presumably because many weeds have ment model with explicit scoring of biolog- redeeming features such as colourful flow- ical, ecological, and geographical attributes ers. In contrast, New Zealand conservation- is a useful biosecurity tool for detecting ists are inclined to consider such taxa as potentially invasive weeds in many areas of weeds. Those with botanical training distin- the world. guish clearly between the native and alien floras, often to the exclusion of the latter in the compilation and collection of botani- Acknowledgement cal data (Raven and Engelhorn, 1971). The interplay between these two floras is a major This work is a largely collaborative effort to issue on oceanic islands (Simberloff, 1995) gather and process information on several hun- dred plant taxa. Mark Boersma, Richard Carter, such as New Zealand, and all alien taxa are David Cooke, Roger Cousens, Jon Dodd, Bryan considered weeds to the extent that they Hacker, Surrey Jacobs, Mark Lonsdale, Michael are perceived to impact on native ecosys- Mulvaney, Dane Panetta, Tom Parnell, Bruce Pen- tems, irrespective of any redeeming features gelly, Darren Phillips, Rod Randall and Craig Walton all contributed to the development of (Williams and Timmins, 1990). The situation the model. Thanks also to the anonymous New could well be different, yet unpredictable, in Zealand experts. other countries. In Argentina and Chile for The Australian Quarantine and Inspection example, some wild adventive taxa have an Service (AQIS) provided funding for resources and technical assistance in Australia and the economic value because of their potential to Ministry of Agriculture and Fisheries in New provide food for poor people (E. Rapoport, per- Zealand. Final preparation of the paper was sonal communication to P.A.W.), and there is funded by the Foundation for Research, Science less inclination to consider them to be weeds. and Technology in New Zealand under contract In contrast to these differing judgments, the (No. 9625). model would be less influenced by the eco- nomic value of a taxon than any group of experts. References The model distinguishes between many useful and non-useful taxa, but some use- Anon. (1994). Agreement on the Application of ful taxa are rejected. This is to be expected, Sanitary and Phytosanitary Measures. Annex 1. World Trade Organization, Geneva. because planned introductions are chosen for Anon. (1995). International Standards for Phy- their ability to survive (Ruesink et al., 1995), tosanitary Measures. Section 1. Guidelines for and the questions asked by the model are Pest Risk Analysis (Draft Standard). Rome: based primarily on biological and ecologi- FAO. cal criteria which identify attributes com- Crawley, M. J. (1987). What makes a community invasible? Colonisation, succession and stability mon to both useful agricultural taxa and (A. J. Gray, M. J. Crawley and M. J. Edwards, weeds (Lonsdale, 1994). These may differ eds), pp. 429–453. Oxford: Blackwell Scientific only in a small number of characteristics Publications. within any single life form (Perrins et al., Drake, J. A., Mooney, H. A., di Castri, F., Groves, R. H., Kruger, F. J., Rejmanek, M. 1992). Any model designed to screen pri- and Williamson, M. (eds) (1989). Biological marily for invasive weeds must reject taxa Invasions: A Global Perspective. Chichester: that are useful yet invasive, e.g. in the New John Wiley.
248 P. C. Pheloung et al. Esler, A. E. (1988). Naturalisation of Plants in ed.), pp. 462–463. Frankston: Weed Science Urban Auckland: A Series of Articles from the Society of Victoria Inc. New Zealand Journal of Botany. Wellington: Raven, P. H. and Engelhorn, T. (1971). A plea for DSIR Publishing. the collection of common plants. New Zealand Esler, A. E., Leifting, L. W. and Champion, P. D. Journal of Botany 9, 217–222. (1993). Biological Success and Weediness of the Reichard, S. E. (1996). Prevention of invasive Noxious Plants of New Zealand. Auckland: MAF plant introductions on national and local lev- Quality Management. els. In Assessment and Management of Plant Forcella, F. and Wood, T. J. (1984). Coloniza- Invasions (J. O. Lucken and J. W. Thieret, eds), tion potentials of alien weeds are related pp. 215–228. New York: Springer. to their native distributions—implications for Reichard, S. E. and Hamilton, C. W. (1997). plant quarantine. Journal of Australian Insti- Predicting invasions of woody plants introduced tute of Agricultural Science 50, 35–41. into north America. Conservation Biology 11, Forcella, F., Wood, T. J. and Dillon, S. P. (1986). 193–203. Characteristics distinguishing invasive weeds Rejmanek, M. (1995). What makes a species within Echium (Bugloss). Weed Research 26, invasive. In Plant Invasions–General Aspects 351–364. and Special Problems (P. Pysek, K. Prach, Groves, R. H. (1986). Plant-invasions of Australia: M. Rejmanek and M. Wade, eds), pp. 3–13. an overview. In Ecology of Biological Invasions: Amsterdam: Academic Publishing. An Australian Perspective (R. H. Groves and Rejmanek, M. and Richardson, D. M. (1996). J. J. Burdon, eds), pp. 1–166. Canberra: Aus- What attributes make some plant species more tralian Academy of Sciences. invasive? Ecology 77, 1655–1661. Halloy, S. (1994). Long term trends in the relative Richardson, D. M., Williams, P. A. and Hobbs, R. J. abundance of New Zealand agricultural plants. (1994). Pine invasions in the southern hemi- Otago Conference Series 2, 125–141. sphere: determinants of spread and invadibility. Hazard, W. H. L. (1988). Introducing crop, pasture Journal of Biogeography 21, 511–527. and ornamental species into Australia—the risk Richardson, D. M., Cowling, R. M., LeMaitre, D. C. of introducing new weeds. Australian Plant (1990). Assessing the risk of invasive success Introduction Review 19, 19–36. in Pinus and Banksia in South African moun- Lonsdale, W. M. (1994). Inviting trouble: intro- tain fynbos. Journal of Vegetation Science 1, duced pasture species in northern Australia. 629–642. Australian Journal of Ecology 19, 345–354. Ruesink, J. L., Parker, I. M., Groom, M. J. and Mack, R. N. (1996). Predicting the identity and Kareiva, P. M. (1995). Reducing the risks of fate of plant invaders: emergent and emerg- nonindigenous species introductions. Bioscience ing approaches. Biological Conservation 78, 45, 465–477. 107–121. Scott, J. K. and Panetta, F. D. (1993). Predicting Noble, I. R. (1988). Attributes of invaders and the Australian weed status of southern African the invading process: terrestrial and vascular plants. Journal of Biogeography 20, 87–93. plants. In Biological Invasions: A Global Per- Simberloff, D. (1995). Why do introduced species spective (J. A. Drake, H. A. Mooney, F. di Castri, appear to devastate islands more than mainland R. H. Groves, F. J. Kruger, M. Rejmanek and areas. Pacific Science 49, 87–97. M. Williamson, eds), pp. 301–310. Chichester: Thompson, K., Hodgson, J. G. and Rich, T. C. G. John Wiley. (1995). Native and alien invasive plants: more Panetta, F. D. (1993). A system of assessing of the same? Ecography 18, 390–402. proposed plant introductions for weed potential. Tucker, K. C. and Richardson, D. M. (1995). An Plant Protection Quarterly 8, 10–14. expert system for screening potentially invasive Panetta, F. D. and Mitchell, N. D. (1991a). Bio- alien plants in South African fynbos. Journal of climatic prediction of the potential distribution Environmental Management 44, 309–338. of some weed species prohibited entry to New Walton, C. and Parnell, T. (1996). Weeds as quar- Zealand. New Zealand Journal of Agricultural antine pests. Proceedings Eleventh Australian Research 34, 341–350. Weeds Conference (R. C. H. Shepared, ed.), Panetta, F. D. and Mitchell, N. D. (1991b). Homo- pp. 462–463. Frankston: Weed Science Society clinal analysis and the prediction of weediness. of Victoria Inc. Weed Research 31, 273–284. Williams, P. A. and Timmins, S. M. (1990). Weeds Parsons, W. T. and Cuthbertson, E. G. (1992). in New Zealand protected natural areas: a Noxious Weeds of Australia. Melbourne: Inkata review for the Department of Conservation. Press. Science and Research Series 14 pp. 1–114. Perrins, J., Williamson, M. and Fitter, A. (1992). Wellington: Department of Conservation. A survey of differing views of weed classifica- Williamson, M. (1993). Invaders, weeds and the tion: implications for regulating introductions. risk from genetically manipulated organisms. Biological Conservation 60, 47–76. Experientia 49, 219–224. Pheloung, P. (1996). Predicting the weed potential Williamson, M. and Fitter, A. (1996). The charac- of plant introductions. In Proceedings, Eleventh teristics of successful invaders. Biological Con- Australian Weeds Conference (R. C. H. Shepard, servation 78, 163–170.
A weed risk assessment model 249 Appendix 1
250 P. C. Pheloung et al.
A weed risk assessment model 251 Appendix 2
You can also read