Section 72 Analysis for 5-Year Review Results Canterbury Regional Pest Management Strategy 2005- 2015

Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005- 2015

Report No. U10/3 ISBN 978-1-877574-29-0 (hard copy) ISBN 978-1-877574-30-6 (electronic)

R Maw

June 2010 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

ii Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Table of Contents

PART 1: INTRODUCTION ...... 1

PART 2: CHILEAN NEEDLE GRASS ...... 3

2.1 BACKGROUND ...... 3 2.2 DISTRIBUTION ...... 3 2.3 IMPACTS ...... 3 2.4 EFFECTS SUMMARY ...... 4 2.5 ANALYSIS ...... 4 2.6 SECTION 72 CONCLUSIONS ...... 8 2.7 FUNDING RATIONALE ...... 8 2.8 ANNEX 1: MAIN ASSUMPTIONS ...... 10 2.9 ANNEX 2: TABLES OF RESULTS ...... 11 2.10 ANNEX 3 CURRENT AND POTENTIAL DISTRIBUTION OF NASSELLA NEESIANA ...... 19

PART 3: PLANTS FOR POSSIBLE INCLUSION IN A RPMS ...... 25

3.1 INTRODUCTION ...... 25 3.2 BACKGROUND ...... 26 3.3 EVALUATION ...... 27 3.4 ACKNOWLEDGEMENTS ...... 99 3,5 REFERENCES ...... 99 3.6 OTHER SOURCES OF INFORMATION: ...... 100

PART 4: ANIMALS FOR POSSIBLE INCLUSION IN A RPMS ...... 103

4.1 EUROPEAN HEDGEHOG (ERINACEUS EUROPAEUS) ...... 103 4.2 SHIP RAT (RATTUS RATTUS), NORWAY RAT (RATTUS NORVEGICUS) ...... 104 4.3 ARGENTINE ANT ...... 105

iii Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

iv Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Part 1: Introduction

Environment Canterbury has undertaken a 5-year review of the Canterbury Regional Pest Management Strategy 2005-2015 (RPMS). Feedback from interested parties suggested that various plants and animals should be considered for inclusion in the RPMS. The Biosecurity Act 1993 prescribes an analysis test be undertaken in accordance with section 72 the Act in order to justify inclusion. The section states: (1) A regional council may notify, in accordance with section 78 of this Act, a proposal for a regional pest management strategy only if it is of the opinion that— (a) The benefits of having a regional pest management strategy in relation to [each organism to which the strategy would apply] outweigh the costs, after taking account of the likely consequences of inaction or alternative courses of action; and (b) The net benefits of regional intervention exceed the net benefits of an individual's intervention; and [(ba) Where funding proposals for the strategy require persons to meet directly the costs of implementing the strategy— (i) The benefits that will accrue to those persons as a group will outweigh the costs; or (ii) Those persons contribute to the creation, continuance, or exacerbation of the problems proposed to be resolved by the strategy; and (c) each organism in respect of which the strategy is under consideration is capable of causing at some time a serious adverse and unintended effect in relation to the region on one or more of the following: (i) Economic wellbeing; or (ii) The viability of threatened species of organisms, the survival and distribution of indigenous plants or animals, or the sustainability of natural and developed ecosystems, ecological processes, and biological diversity; or (iii) Soil resources or water quality; or (iv) Human health or enjoyment of the recreational value of the natural environment; or (v) The relationship of Maori and their culture and traditions with their ancestral lands, waters, sites, waahi tapu, and taonga.

The information included in Parts 2-4 is a compilation of the material used to guide Council’s opinion regarding the proposed inclusion of Chilean needle grass in the RPMS and the changes to the site-led biodiversity programmes. The information is the work of the authors who have produced the reports, papers or publications referenced.

1 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

2 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Part 2: Chilean Needle Grass

Environment Canterbury commissioned the following report into Chilean needle grass. Meeting the requirements of the biosecurity act 1993: economic evaluation of regional pest management strategy for plant pests, Harris Consulting May 2010.

2.1 Background Chilean needle grass is an erect, tufted perennial grass, which can grow up to one metre high in the absence of grazing. It originates from South America, and has naturalized in New Zealand in Hawke’s Bay, Marlborough, and Auckland. Plants form dense clumps, which exclude preferred pasture species and are unpalatable to stock during the flowering period. Chilean needle grass flowers between November and April and produces sharp tipped seeds, which can bore into the eyes and pelts of grazing animals. The seeds can be moved by stock, waterways, feral animals, machinery, hay, grain and to some extent, by wind.

2.2 Distribution Chilean needle grass is recognised as a weed of national significance in Australia. In New Zealand, there are localised infestations in Auckland and Hawke’s Bay in the North Island and more extensive infestations in Marlborough. Until recently Canterbury was thought to be free of Chilean needle grass. However an infestation was discovered recently in a vineyard in Spotswood, and the current infestation is estimated to be approximately 80 ha, with a nil to isolated plant infestation across approximately 95% of the area and 5% scattered to dense.

Marlborough is the likely source of the infestation in Canterbury. In Marlborough, ninety-six properties are known to have an infestation of Chilean needle grass in 2005. Infestations there range from isolated patches to widespread infestations and cover an estimated area of 4300 hectares (Bell, 2005)1. The areas in Marlborough increased from ~1500 ha in 1987 to 4300 ha in 2005, showing its potential for rapid increase and spread.

Table 1: Areas of Chilean needle grass infestation in Marlborough, 2005 survey

Classification Area (ha, 2005) Fringe (<5% ground cover of CNG2) 1346 (31%) Core (5-50% ground cover of CNG 2106 (49%) Nucleus (>50% ground cover CNG) 859 (20%) Total 4311 (100%)

2.3 Impacts The impacts of Chilean Needle grass are summarised in the table below. Economic Impacts Conservation Values Soil resources or water Human Health or quality recreational values

Current Potential Current Potential Current Potential Current Potential

Low High Nil Medium Nil Nil Nil Nil

1 Bell, M.D. 2006 “Spread of Chilean Needlegrass (Nasella neesiana) in Marlborough, New Zealand”. NZ Plant Protection 59:266 – 270 (2006)

3 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

The Esler2 weediness rating for Chilean needlgrass is 11/40, and the Biological Success and Environmental Impact Rating is 18/60, which indicates a reasonably significant weed potential.

The economic impacts of Chilean needle grass are likely to occur to pastoral landholders throughout Canterbury over time from exclusion of palatable grazing species and disruption to the farm operations. Typically farmers have to keep stock of infested paddocks for a period of 3 months in summer while the plant seeds. Chilean needle grass is a difficult weed to control and is not readily killed by approved herbicides.

Chilean needle grass also impacts on animal welfare, as sheep, cattle and dogs can be affected by the sharp seeds. These are known to burrow through skin and into muscle tissue, entering eyes, and causing severe pain and infection. This can also cause economic damage through downgrading of pelts.

Chilean needle grass has an impact on biodiversity through exclusion. It has been shown to form very dense stands, excluding all other material, thus reducing biodiversity in affected areas3.

2.4 Effects summary Chilean needle grass appears capable of causing damage to Canterbury’s pastoral farming economy. Therefore, a Regional Pest Management Strategy in respect of this pest will satisfy the requirements of Section 72(c) Part (i) of the Biosecurity Act 1993. Environment Canterbury is recommending that Chilean needle grass is a ‘containment control’ plant pest in the proposed Regional Pest Management Strategy.

2.5 Analysis The analysis compares two scenarios – the situation with No RPMS in place, and the situation with a RPMS. These two scenarios are discussed below, and compared in the following section.

2.5.1 Scenario 1: No RPMS In this scenario no control of Chilean Needle grass is undertaken, and the plant spreads throughout the region. AgResearch has estimated the potential habitat of Chilean needle grass in Canterbury using the CLIMEX model4. They estimates that 1.2 million ha is suitable or optimal habitat for Chilean needle grass, of which approximately 43% is in Land Use Capability classes 1 – 3 (highly versatile) and 53% in less versatile land classes 4 – 6, and the remainder in classes 7 and 8.

2The Esler rating is an assessment of a number of characteristics likely to indicate the weediness of a plant. It is based on a scale from 1 to 24. (Esler, 1994) 3 Anonymous (2003). Weed Management Guide - Chilean needle grass (Nassella neesiana). CRC for Weed Management. Cited in Bourdot et al 2010. “Current and potential distributions of Nasella neesiana (Chilean needle grass) in Australia and New Zealand. 17th Australasian Weeds conference, 26 – 30 September 2010. Christchurch NZ. 4 G. Bourdot. AgResearch. Pers.comm. 2010.

4 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Table 2: Estimates of potential habitat (class Suitable orOptimal) for Chilean needle grass in Canterbury (Bourdot and Lamoureaux, 2010 pers comm.)

Pasture type LUC class High producing Low producing Total 1 to 3 524,005 5,040 529,044 4 to 6 489,987 161,382 651,368 7,8 14,646 23,819 38,466 Total 1,028,637 190,241 1,218,878

In order to estimate the costs of Chilean needle grass spreading from its current habitat, a model of plant growth and infestation was used to determine the outcome of no regional intervention. The model used has three elements: • the weed increases to saturation in an already infested area; • it infests new areas; and • it establishes at new locations.

Pest Increase to Saturation

A growth curve is used to model the rate at which the pest increases to saturation, where the density of the plants (plants/ha) increases in the fashion:

Increase in pest density Pest density (%)

Years

The shape of the curve with which the plant density increases within an infested area is defined within the model using a theta logistic equation: Nt=N0.K/[N0+(K-N0).exp(-r.t)],

The model estimates a value of r that fits the time (t) taken to increase from current density (No) to maximum density (K). This value of r is estimated using the following formula:

r =1/t.ln{(K-N0)/[(1/0.99-1).N0]}

The model was set up to approximately represent the rate of increase that occurred in Marlborough between surveys of Chilean needle grass infestations in 1987 and 2005 surveys. This utilised an increase in density in the 0 – 5% (2.5% midpoint used) density class to 50% which occurred over ~30 years. Modeling this rate of increase resulted in an r value of 0.118 – 0.178. This value will be affected by the fact that:

5 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

• Not all of the 0 – 5% density class in Marlborough exceeded 50% density after 30 years • Control was undertaken on those properties for much of the interval.

Nevertheless it was considered that this represented adequately reasonable boundaries of the potential density increase. These rates of increase were applied in the model to the Canterbury situation. Initial densities of existing areas infested in Canterbury are assumed to be 2%. Newly infested areas have an assumed density of 0.5%, and increase in density as per the theta logistic curve discussed above. Each cell of infestation is increased in density separately.

Infestation of new areas

In the model, plants infest an area, become established, then infest new areas. It is assumed that plants on the boundary of infested areas will spread a certain distance into uninfested areas. The maximum distance of spread is used to define the size of the newly infested area, and the new infestation is assumed to occur as an increasing rectangle, the size of which is defined initially by the user. The user also sets the nearest boundaries that restrict spread, and the maximum width of the rectangle.

For Canterbury the model was set up to represent the increase in area that occurred in Marlborough from 1500 ha to ~4000 ha between 1987 and 2005. This is approximated by spread values of 15 – 30m. Gardner et al (2003)5 report that typical immediate seed spread is less than 2.8 m, and therefore these spread distance values are likely to represent more than simple seed drop, and reflect the fact that some local animal mediated spread is likely to occur.

Infestation of new sites

As well as increasing the area of existing sites the model establishes new locations of infestation. The new locations are established as new cells independent of the original infestation, with initial densities as per new spread. The sites are established as a proportion of the existing area infested – which assumes that the greater the seed availability, the greater the potential for establishment of new sites. The frequency of establishment of new sites, and the number of new sites established per ha of existing infestation, is based on the number of new properties infested in Marlborough over the period 1987 – 2005. This works out to approximately 1 new property infested every year for each 565 ha of initial infestation.

Financial Assumptions

Control costs are included in the model. 65% of properties are assumed to carry out control, which comprises all the land in classes 1 – 3 (45% of total potential area) where it is assumed to be highly economic to control, and the remainder in classes 4 – 6 (20%) where it is less economic to control. Control is either complete or is not conducted at all, but it is assumed that, in the absence of regional intervention, landholder control will not be sufficiently effective to prevent the further spread of the plant. Control costs are estimated at $140/ha/annum control costs (spot spray, labour 3 hours/ha @$30/hour and chemical $50/ha). This is roughly comparable with control costs reported by farmers in infested areas of Marlborough of $10,000 annually to control 120/ha, and $1000 to control 2ha of boundary

5 Gardener, M.R., Welly, R.D.B. and Sundel, B.M. 2003 “Ecology of Nasella neesiana, Chilean needle grass, in pastures on the Northern Tablelands of New South Wales. I: Seed production and dispersal.” Australian Journal of Agricultural Research 2003 54:613 - 619

6 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

by helicopter. These control costs are significantly below the costs experienced by ECan in its control of the weed in Canterbury so far. Control costs may decrease with the introduction of Taskforce which can be broadcast and has residual action, and this is tested in the analysis by assuming that control is required only every 3 years instead of annually. Loss of production is estimated using farmer interviews and work undertaken by Hunter (2001)6 The financial costs for farm operations are based on the weighted average production for the land uses where control is not undertaken, from the MAF Farm Monitoring Canterbury Hill Country model for the 2007 – 08 years (it is assumed that no loss occurs on the more versatile soils where control is undertaken). The loss of production is linearly related to the displacement of desirable species by the weed, which is assumed to be 20% less productive – a total of $30/ha. Where Chilean needle grass densities exceed 5%, there is an additional loss equal to revenue from 3 months lack of grazing over summer – estimated at $73/ha.

No RPMS scenario outcome

The modeling indicates that in the absence of regional control Chilean needle grass will cost a NPV of between $0.4 and $1.0 million. Under the modeling taken here the majority of the cost is incurred as a result of landholder control, which is assumed to take place on the higher value pastoral land and a proportion of the lower value pastoral land. Testing the assumption about control costs on the basis that Taskforce (a needle grass specific herbicide with residual action) was permitted in New Zealand reduces the cost of the No RPM scenario to between $0.1 and $0.36 million.

In addition there are likely to be: • significant disruption to farming systems to avoid needle grass during seeding time of year; • animal welfare concerns from harm caused by the seeds to stock, dogs and other animals; and • loss of biodiversity from exclusion by Chilean needle grass.

2.5.2 Scenario 2: RPMS7 In this option a containment strategy is adopted. This involves complete control of all land currently identified as having Chilean needle grass infestations, and searching for any new infestations in the surrounding areas. The cost of the RPMS is estimated at $45,000 for the first year, $35,000 for the second year, and $15,000 per annum thereafter8. Of this approximately 33% is control and the remainder inspection costs. This represents a NPV of $230,000 in total to control the pest.

Because the requirements for expenditure beyond year 5 are uncertain, the model was sensitivity tested with costs of $35,000 ongoing beyond year 5.

6 Hunter, R. 2001. “Financial, Technical and Social Impacts on Farming within the Marlborough District Council Region.” Report Prepared for the Marlborough District Council. 7 Only one RPMS scenario is considered here because eradication has been attempted and found to be not technically feasible. 8 Ray Maw, ECan, pers.comm.

7 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

2.6 Section 72 Conclusions 2.6.1 Section 72(a) Control of Chilean needle grass will prevent damage to over 1.2 million ha of pastoral habitat. Modeling indicates that under all the main scenarios of spread and infestation the control of Chilean Needle grass in a RPMS produces a positive net benefit of between $0.2 and $0.8 million NPV. The level of benefit may reduce if Taskforce becomes available and control costs reduce, but at the higher levels of spread there would still be benefit in the RPMS. At higher regional control costs the strategy still shows a net benefit of between $0.1 and 0.7 million under the main scenario. In addition to the financial benefits prevention of damage to animal welfare and biodiversity are benefits that should be taken into account.

If the council is satisfied that the assumptions used in modeling the spread of Chilean needle grass are reasonable or if control costs are likely to be lower, but higher levels of spread are likely to be experienced, then the requirements of Section 72(a) will have been met.

2.6.2 Section 72(b) Part of the value protected by control of Chilean needle grass are regional values through prevention of spillover from currently infested areas to clear areas. The amount of regional damage from spillover to clear properties will amount to between $0.3 and $0.9 million, depending on the spread assumptions used. This compares with a NPV of $0.2 million for the regional costs of inspection, monitoring etc. There is likely therefore to be a surplus of $0.1 to $0.6 million in regional benefit under the main assumptions modeled.

If the costs of the strategy prove to be higher than expected from year 5 onward (tested at $35,000 per annum vs $15,000 per annum in the main scenario), the costs of the RPMS could be as high as $0.3 million. Under this scenario the net regional benefit of the strategy would vary from approximately $0 to $0.8 million.

The requirements of Section 72(b) are therefore likely to be met under all but the more conservative assumptions regarding Chilean needle grass spread and costs of strategy implementation.

2.6.3 Section 72(ba) The values protected by the control of Chilean needle grass are largely pastoral production values. A charge against rural landholders will therefore satisfy the requirements of Section 72(ba). A charge for control costs against landholders on whose properties Chilean needle grass is currently located will also satisfy Section 72(ba) on the basis that these individuals are exacerbators. A charge to the regional community for part of the costs of the strategy will satisfy the requirements of Section 72(ba) on the basis of animal welfare and biodiversity benefits to the wider community.

2.7 Funding Rationale In terms of funding, the rationale for regional intervention is primarily related to the production benefits and reduced control costs for landholders who would be potentially affected by Chilean needle grass and it is appropriate therefore that these parties are charged the majority of the costs for the RMPS. However it should be noted that there are also benefits to the wider regional community from reduced impacts to animal welfare and a reduction in affects to biodiversity. In the long term the impact to biodiversity could be quite significant as 190,000 ha of low producing grassland is suitable or optimal habitat for Chilean needle grass. A proportion charge to the regional community may be appropriate, but it should be

8 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

noted that the ratio of high producing to low producing grassland likely to be affected is 5.5:1. As the biodiversity values are more likely to be on the lower producing grassland, the charge to the regional community should reflect the predominance of the productive benefits of the strategy. The proportion of the strategy charged to the regional community would be difficult to justify above 25% - 50%.

Table 3: Analysis of funding options for Chilean needle grass

Pest Cost Item Beneficiaries Limits of acceptable charge Range of Comment potential charge Chilean Inspection • Landholders from • Rural landholders receive • Rural A Works and Services Needle prevention of spillover benefit greater than 100% of landholders 0 across rural landholders grass • General community for the cost of compliance – 100% is the recommended prevention of damage inspection • Region 0% - option as it charges all to conservation values • Potential for damage to 50% beneficiaries directly. associated with dry biodiversity up to 180,000 A charge to the general grassland ha. Damage to animal community may be communities. welfare not able to be appropriate up to 25% - • General community defined. Benefits to regional 50% at the extreme for from prevention of community not able to be prevention of damage damage to animal well defined but unlikely to to community values. welfare be greater than 25% - 50% of benefit. Control • Landholders from • Landholders with Chilean Landholders with Exacerbator pays is not prevention of spillover needle grass can be Chilean needle recommended to low charged as exacerbators grass 0% - 100% level pests where • General community for • Rural landholders receive complete containment prevention of damage benefit in excess of control Rural landholders is required since it to conservation values costs 0 – 100% places onerous associated with dry • Likely to be lower cost to demands on grassland general community from Region 0% - 50% landholders, and communities. control at low levels even complete control is where community values are unlikely to be achieved. not directly threatened. • General community Charge as for the from prevention of inspection costs. damage to animal welfare

9 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

2.8 Annex 1: Main Assumptions

Chilean needle Plant grass Current Area infested 80 ha Number of sites infested currently 1 Maximum Area potentially infested 1,180,413 ha Current densities 2% Density of new infestations 0.50% Maximum density 70% Years to significant seed spread 2 Distance of seed spread (Min) 15m Distance of seed spread (Max) 30m Control Costs for Periodic Control $150/annum Control Interval for Periodic Control 1 year Proportion of properties undertaking periodic control 65% Other losses ($/ha) $73/ha Gross Margin of agricultural production benefit $30/ha

10 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

2.9 Annex 2: Tables of Results

Table 4: Model outputs for main assumptions Chilean needle grass (NPV 8%, 100 years)

Pest Model Output Pest : Chilean needle grass Proportion of landholders controlling : 65%

Years 1 2 3 4 5 6 7 8 9 10 Years to 90% of max coverage = 40; Spread Distance = 15; R = 0.1782925 Area Infested 80 88 90 98 101 110 114 125 129 142 Area Displaced by Pest 1 1 1 1 1 1 2 2 2 2 Area regularly controlled by Landholders 52 57 58 64 65 72 74 81 84 92 Cost of Lost Production $90 $31 $42 $82 $105 $161 $847 $849 $852 $855 Cost of Landholder Control $7,800 $8,595 $8,760 $9,559 $9,809 $10,729 $11,082 $12,140 $12,613 $13,827 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 NPV of No Strategy $381,124 NPV of Strategy $232,339 Net Benefit of Strategy $148,785 Loss in initially infested area $103,929 Net Regional Benefit $122,331

Years to 90% of max coverage = 40; Spread Distance = 30; R = 0.1782925 Area Infested 80 95 97 114 119 139 147 172 183 213 Area Displaced by Pest 1 1 1 1 1 1 2 2 2 3

Area regularly controlled by Landholders 52 62 63 74 77 90 96 112 119 139 Cost of Lost Production $90 $33 $44 $87 $111 $172 $853 $859 $864 $872 Cost of Landholder Control $7,800 $9,229 $9,479 $11,104 $11,576 $13,565 $14,331 $16,754 $17,886 $20,815 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000

11 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

NPV of No Strategy $997,092 NPV of Strategy $232,339 Net Benefit of Strategy $764,753 Loss in initially infested area $103,929 Net Regional Benefit $738,299

Years to 90% of max coverage = 60; Spread Distance = 15; R = 0.1188616

Area Infested 80 88 90 98 101 110 114 125 129 142 Area Displaced by Pest 1 1 1 1 1 1 1 1 1 2 Area regularly controlled by Landholders 52 57 58 64 65 72 74 81 84 92 Cost of Lost Production $97 $34 $46 $90 $115 $177 $19 $27 $33 $852 Cost of Landholder Control $7,800 $8,595 $8,760 $9,559 $9,809 $10,729 $11,082 $12,140 $12,613 $13,827 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 NPV of No Strategy $370,585 NPV of Strategy $232,339 Net Benefit of Strategy $138,245 Loss in initially infested area $101,600 Net Regional Benefit $114,120

Years to 90% of max coverage = 60; Spread Distance = 30; R = 0.1188616 Area Infested 80 95 97 114 119 139 147 172 183 213 Area Displaced by Pest 1 1 1 1 1 1 1 1 2 2 Area regularly controlled by Landholders 52 62 63 74 77 90 96 112 119 139 Cost of Lost Production $97 $36 $48 $95 $121 $186 $31 $43 $52 $865 Cost of Landholder Control $7,800 $9,229 $9,479 $11,103 $11,576 $13,564 $14,330 $16,752 $17,882 $20,809 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 NPV of No Strategy $929,707 NPV of Strategy $232,339 Net Benefit of Strategy $697,368 Loss in initially infested area $101,600 Net Regional Benefit $673,242

12 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Table 5: Model outputs for Chilean needle grass with low cost control option (3 year spray) (NPV 8%, 100 years)

Pest Model Output Pest : Chilean needle grass Proportion of landholders controlling : 65% Years 1 2 3 4 5 6 7 8 9 10 Years to 90% of max coverage = 40; Spread Distance = 15; R = 0.1782925

Area Infested 80 88 90 98 101 110 114 125 129 142 Area Displaced by Pest 1 1 1 1 1 1 2 2 2 2 Area regularly controlled by Landholders 52 57 58 64 65 72 74 81 84 92 Cost of Lost Production $90 $37 $50 $97 $124 $191 $847 $849 $852 $855 Cost of Landholder Control $2,600 $2,865 $2,920 $3,186 $3,270 $3,576 $3,694 $4,047 $4,204 $4,609 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 NPV of No Strategy $138,008 NPV of Strategy $232,339 Net Benefit of Strategy -$94,331 Loss in initially infested area $38,959 Net Regional Benefit -$55,815

Years to 90% of max coverage = 40; Spread Distance = 30; R = 0.1782925 Area Infested 80 95 97 114 119 139 147 172 183 213 Area Displaced by Pest 1 1 1 1 1 1 2 2 2 3 Area regularly controlled by Landholders 52 62 63 74 77 90 96 112 119 139 Cost of Lost Production $90 $39 $52 $103 $131 $202 $853 $859 $864 $872 Cost of Landholder Control $2,600 $3,076 $3,160 $3,701 $3,859 $4,522 $4,777 $5,585 $5,962 $6,938 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 NPV of No Strategy $356,820

13 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Pest Model Output Pest : Chilean needle grass Proportion of landholders controlling : 65% Years 1 2 3 4 5 6 7 8 9 10 NPV of Strategy $232,339 Net Benefit of Strategy $124,481 Loss in initially infested area $38,959 Net Regional Benefit $162,997

Years to 90% of max coverage = 60; Spread Distance = 15; R = 0.1188616 Area Infested 80 88 90 98 101 110 114 125 129 142 Area Displaced by Pest 1 1 1 1 1 1 1 1 1 2 Area regularly controlled by Landholders 52 57 58 64 65 72 74 81 84 92 Cost of Lost Production $97 $40 $54 $105 $134 $206 $19 $27 $33 $852 Cost of Landholder Control $2,600 $2,865 $2,920 $3,186 $3,270 $3,576 $3,694 $4,047 $4,204 $4,609 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 NPV of No Strategy $131,104 NPV of Strategy $232,339 Net Benefit of Strategy -$101,235 Loss in initially infested area $36,630 Net Regional Benefit -$60,390

14 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Pest Model Output Pest : Chilean needle grass Proportion of landholders controlling : 65% Years 1 2 3 4 5 6 7 8 9 10 Cost of Landholder Control $2,600 $3,076 $3,160 $3,701 $3,859 $4,521 $4,777 $5,584 $5,961 $6,936 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 $15,000 NPV of No Strategy $324,990 NPV of Strategy $232,339 Net Benefit of Strategy $92,651 Loss in initially infested area $36,630 Net Regional Benefit $133,496

15 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Table 6: Model outputs for high RPMS cost assumptions Chilean needle grass (NPV 8%, 100 years)

Pest Model Output Pest : Chilean needle grass Proportion of landholders controlling : 65%

Years 1 2 3 4 5 6 7 8 9 10

Years to 90% of max coverage = 40; Spread Distance = 15; R = 0.1782925 Area Infested 80 88 90 98 101 110 114 124 129 142 Area Displaced by Pest 1 1 1 1 1 1 2 2 2 2 Area regularly controlled by Landholders 52 57 58 64 65 72 74 81 84 92 Cost of Lost Production $90 $1 $1 $3 $4 $5 $847 $849 $852 $855 Cost of Landholder Control $7,800 $8,594 $8,759 $9,559 $9,809 $10,728 $11,081 $12,139 $12,612 $13,826 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $35,000 $35,000 $35,000 $35,000 $35,000 NPV of No Strategy $378,864 NPV of Strategy $286,687 Net Benefit of Strategy $92,178 Loss in initially infested area $103,929 Net Regional Benefit $92,897

Years to 90% of max coverage = 40; Spread Distance = 30; R = 0.1782925 Area Infested 80 95 97 114 119 139 147 172 183 213 Area Displaced by Pest 1 1 1 1 1 1 2 2 2 3 Area regularly controlled by Landholders 52 62 63 74 77 90 96 112 119 139 Cost of Lost Production $90 $3 $4 $8 $10 $16 $853 $859 $864 $872 Cost of Landholder Control $7,800 $9,229 $9,479 $11,104 $11,576 $13,565 $14,331 $16,754 $17,886 $20,815 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $35,000 $35,000 $35,000 $35,000 $35,000 NPV of No Strategy $996,783 NPV of Strategy $286,687

16 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Pest Model Output Pest : Chilean needle grass Proportion of landholders controlling : 65%

Years 1 2 3 4 5 6 7 8 9 10 Net Benefit of Strategy $710,096 Loss in initially infested area $103,929 Net Regional Benefit $710,816

Years to 90% of max coverage = 60; Spread Distance = 15; R = 0.1188616 Area Infested 80 88 90 98 101 110 114 125 129 142 Area Displaced by Pest 1 1 1 1 1 1 1 1 1 2 Area regularly controlled by Landholders 52 57 58 64 65 72 74 81 84 92 Cost of Lost Production $97 $4 $5 $10 $13 $21 $19 $27 $33 $852 Cost of Landholder Control $7,800 $8,595 $8,760 $9,559 $9,809 $10,729 $11,082 $12,140 $12,613 $13,827 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $35,000 $35,000 $35,000 $35,000 $35,000 NPV of No Strategy $370,301 NPV of Strategy $286,687 Net Benefit of Strategy $83,614 Loss in initially infested area $101,600 Net Regional Benefit $86,663

17 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Pest Model Output Pest : Chilean needle grass Proportion of landholders controlling : 65%

Years 1 2 3 4 5 6 7 8 9 10 Production Cost of Landholder Control $7,800 $9,229 $9,479 $11,103 $11,576 $13,564 $14,330 $16,752 $17,882 $20,809 Strategy Costs $45,000 $35,000 $15,000 $15,000 $15,000 $35,000 $35,000 $35,000 $35,000 $35,000 NPV of No Strategy $929,423 NPV of Strategy $286,687 Net Benefit of Strategy $642,737 Loss in initially infested area $101,600 Net Regional Benefit $645,785

18 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

2.10 Annex 3 Current and potential distribution of Nassella neesiana

19 For 17th Australasian Weeds Conference, 26-30 Sept 2010, Christchurch, NZ

Current and potential distributions of Nassella neesiana (Chilean needle grass) in Australia and New Zealand

Graeme W. Bourdôt1, Shona L. Lamoureaux1, Darren J. Kriticos2, Michael S. Watt3 and Matthew Brown4 1AgResearch, Lincoln, Private Bag 4749, Christchurch 8140, New Zealand Email: [email protected] 2CSIRO Entomology, GPO Box 1700, Canberra, ACT, 2601 Australia 3Scion, PO Box 29237, Fendalton, Christchurch, New Zealand 4AgResearch, Invermay, Private Bag 50034, Mosgiel 9053, New Zealand

Summary Nassella neesiana (Trin. & Rupr.) In both Australia and New Zealand N. neesiana Barkworth, var neesiana (Chilean needle grass), is is the subject of community-level management an invasive weed in Australia and New Zealand initiatives aimed at local control and prevention of where it is the subject of management programmes spread. It is a “Weed of National Significance” to reduce its impacts (downgrading of wool, skins, (WONS) in Australia (Snell et al. 2007) and is a hides and carcasses, reduced stock carrying prohibited species under the Quarantine Act 1908, capacity, reduced grassland biodiversity) and preventing its sale and distribution. In the ACT and spread. Inferring the species’ climate preference parts of NSW it is a “Declared Pest Plant” requiring from its distribution in its native range in South its control by landholders (Anonymous 2003). In America using CLIMEX, we estimate that 180 and New Zealand N. neesiana is a “Total Control Plant” 15 million ha respectively are climatically suitable in Hawke’s Bay (HBRC 2009) and a “Containment in Australia and New Zealand under current climate. Plant” in Marlborough (MDC 2009) requiring We also estimate that 0.24 and 0.52% respectively landholders to eradicate and contain the species of this suitable area has been invaded in Australia respectively. Two assumptions underpinning these and New Zealand. These results imply that N. measures are that the species has not yet realised its neesiana could become a much greater problem in potential range and therefore its potential ecological both Australia and New Zealand and that or economic impact in either Australia or New management to limit its spread is justified. Zealand, and that without control it will spread to occupy more of its potential range in both countries. Keywords Climate, CLIMEX, niche model, Here we test the first of these assumptions by weed management. firstly defining the potential geographic ranges of N. neesiana in Australia and New Zealand and INTRODUCTION secondly by comparing the size of each of these Nassella neesiana (Trin. & Rupr.) Barkworth, var potential ranges with the size of their invaded parts. neesiana [synonym Stipa neesiana] or Chilean needle grass (family Gramineae; sub-family MATERIALS AND METHODS Pooideae; tribe Stipeae) is a tufted perennial grass CLIMEX version 3 (Sutherst et al. 2007), a of temperate South America origin. It has dynamic climate model integrating weekly growth naturalised in both Australia and New Zealand, and survival (stress) responses of a species to being first recorded in Australia in 1935, in temperature and soil moisture into an annual index Melbourne (McLaren et al. 1998), and in Auckland of climatic suitability, the Ecoclimatic Index (EI) in New Zealand sometime before 1940 (Bourdôt (ranging from 0 for locations where the species and Hurrell 1989). It reduces the livestock carrying cannot persist to 100 for optimal locations) was capacity of pastures due to the production of masses parameterised for N. neesiana. The parameters of unpalatable flower stalks (Anonymous 2003; (Table 1) were fitted to the species’ native and Gardener et al. 2003) and its sharp penetrating seeds introduced ranges in South America by iteratively injure livestock and result in the downgrading of changing their values (informed by published wool, skins, hides and carcasses (Bourdôt and Ryde literature and anecdote) until the model’s projected 1986). The weed also reduces the biodiversity of distribution of EI closely corresponded to the 90 native grasslands in Australia (Anonymous 2003). known occurrences in South America. The draft model was verified by projecting it onto the UK and 2

Western Europe. This comparison revealed that the Table 1. Values of the CLIMEX model parameters model predicted EI≥1 for all but three occurrences (Sutherst et al. 2007) fitted for Nassella neesiana. (all in Scotland). By reducing the tolerable length of Index Parameter Value Units the growing season (PDD) in the model from 900 to Growth 650 0C days, these three points were encompassed Temp. Lower threshold 8 0C with a slight, but ecologically reasonable increase in Lower optimum 20 0C the suitable area in South America. This model was Upper optimum 25 0C then used without further modification to project the Upper threshold 28 0C species’ potential distribution in Australia and New Moisture Lower threshold 0.1 Zealand where it was validated by comparison with Lower optimum 0.7 all known occurrences. Upper optimum 1.1 A 0.50 of arc (ca. 50 x 50 km) climate dataset Upper threshold 1.3 generated by Kriticos et al. (2006) from the 1961- Stresses 1990 climate normals provided by the Climatic Cold Threshold 0 0C Research Unit, University of East Anglia (described Accumulation rate -0.01 Wk-1 by New et al. 1999) was used to construct the Heat Threshold 33 0C model. Finer-scale climate data sets (0.050arc, ca. 5 Accumulation rate 0.005 Wk-1 x 5 km) used to project the model onto Australia Dry Threshold 0.1 and New Zealand were generated by Kriticos (2010) Accumulation rate -0.02 Wk-1 and by Kriticos using data from Leathwick and Wet Threshold 1.3 Stephens (1998) respectively. Accumulation rate 0.002 Wk-1 The percentage of the climatically suitable land H-W Temp. threshold 25 0C area infested by N. neesiana in Australia and in Moist. threshold 1.2 New Zealand was calculated using a GIS as the sum Accumulation rate 0.01 Wk-1 of the land areas of the 0.050arc climate cells with Growing Degree-day threshold 650 0C EI≥1 that contained one or more occurrences of N. season for persistence days neesiana divided by the total land area of all of the 0 0.05 arc cells with EI≥1 in each of the countries. In Australia, large tracts of land in the south Climate grid cells were clipped to fine-scale west of Western Australia are climatically suitable coastlines prior to summarizing the areas of climate as are parts of south-eastern Queensland, regions habitat suitability. from which the species is currently unknown, apart from two occurrences in southern Queensland (Fig. RESULTS 1). The model also suggests that N. neesiana could The parameters for the CLIMEX model for N. naturalise further north in these regions of Australia neesiana are in Table 1. The inferred optimal than claimed in a previous study (McLaren et al. 0 temperature for population growth is 20-25 C and 2004). The model projects that there are 180 million the optimal soil moisture is 0.7-1.1 (70-110%) field ha (1.8 m km2) in Australia with EI≥1.0, and that capacity. In addition, N. neesiana is inferred to 432,157 ha (4,322 km2) are currently occupied by 0 accumulate cold stress at temperatures below 0.0 C, N. neesiana. Thus it is estimated that only 0.24% 0 heat stress above 33 C, dry stress at soil moisture (0.432/180*100) of the climatically suitable land levels below 0.1, wet stress above 1.3, and hot-wet area in Australia has been invaded to date (Table 2). stress when temperature and soil moisture exceed In New Zealand also, large tracts of land beyond 0 25 C and 1.2 field capacity respectively (Table 1). the currently invaded areas are projected to be These parameter values imply that N. neesiana has a climatically suitable (Fig. 2). In the North Island, wide ecological amplitude, tolerating drought-prone large parts of the regions of Northland, Auckland, and seasonally waterlogged soils, supporting field Waikato, Gisborne, Hawkes Bay, Manawatu- observations to this effect (McLaren et al. 1998). Wanganui and Wellington are climatically suitable. This model, when projected onto Australia and In the South Island, large parts of the Nelson, New Zealand, reveals that N. neesiana is potentially Marlborough and Tasman regions are climatically able to naturalise in both countries over geographic suitable, as are eastern Canterbury, eastern Otago ranges that greatly exceed the known current and much of Southland. Only a small fraction of this distributions of the species (Fig. 1 & 2). climatically suitable area in New Zealand has been invaded. The model projects that there are 15 million ha (149,916 km2) in New Zealand with spikelets (fruits) tangling and dropping to the EI≥1.0, and that 78,173 ha (782 km2) are currently ground in a mass near the parent plant rather than occupied by N. neesiana. Thus it is estimated that dispersing away from the originating panicle. By that only 0.52% (0.078/15*100) of the climatically contrast, long-distance human-mediated dispersal of suitable land area in New Zealand has been invaded the seeds of N. neesiana appears to have driven the to date (Table 2). invasion of this species in its exotic ranges. Its occurrences are commonly associated with tanneries and the transport of animals and/or their hides or fleeces (Haywood and Druce 1919; Snell et al. 2007; Stace 2001). Mechanical control of roadside populations and use of earthmoving machines are implicated in its spread in Australia (Anonymous 2003). As a result, programmes that prevent the transport of animals, hides and fleeces from infested areas to the climatically suitable areas projected by this CLIMEX model can be expected to limit the spread of the species and thereby reduce its future impacts. Similarly, adherence to strict hygiene measures with respect to machinery used in N. neesiana-infested areas such as roadsides and sports fields can be expected to reduce the risk of spread. To this end, regionally or nationally-coordinated management programmes such as the WONS in Figure 1. Potential distribution and known Australia may be justified in New Zealand. occurrences of N. neesiana in Australia.

Table 2. Comparison of the land area (million ha) climatically suitable for Nassella neesiana (EI≥1) with the land area invaded. Country Suitable Invaded % area area invaded Australia 180 0.432 0.24 New Zealand 15 0.078 0.52

DISCUSSION The CLIMEX model for Nassella neesiana presented here, in combination with the known occurrences of the species in Australia and New Zealand, reveals that it has occupied less than 1% of the land areas that are currently climatically suitable in these two countries. This result indicates that this weed, approximately 70 years after being first recorded as naturalised (McLaren et al. 1998) (Bourdôt and Hurrell 1989), remains in the early stages of its invasion in both countries. Therefore much wider geographic distributions, and hence much greater ecological and economic impacts, are possible in the future. The realisation of these projected future impacts will depend upon the extent to which the propagules of the species are dispersed to climatically suitable areas. The natural dispersal of this species by wind appears to be limited by the bigeniculate awn Figure 2. Potential distribution and known (Conner et al. 1993) that results in the mature occurrences of N. neesiana in New Zealand. 4

In New Zealand, despite the current production and dispersal. Australian Journal of management programmes in Hawke’s Bay and Agricultural Research 54, 613-619. Marlborough, local scale, farm to farm spread of N. Haywood, I.M. and Druce, G.C. (1919) 'The neesiana is ongoing. Evidence of this is apparent in adventive flora of Tweedside' (Buncle & Co., Marlborough where the number of farms known to Arbroath). support populations of the weed has increased HBRC (2009). Total control plant pests. exponentially from 18 in 1987, to 96 in 2005 (Bell http://www.hbrc.govt.nz/WhatWeDo/Pests/Plan 2006). Additionally, the discovery of N. neesiana in ts/TotalControlPlantPests/tabid/170/Default.asp Canterbury in 2008 apparently originating as seeds x. 14 September. on livestock transported from Marlborough, ca. 200 Kriticos, D.J., Alexander, N.S. and Kolomeitz, S.M. km away (Laurence Smith, ECan, pers. comm.) (2006). Predicting the potential geographic (Fig. 2) provides evidence that long distance distribution of weeds in 2080. Fifteenth human-mediated dispersal is occurring. This recent Australian Weeds Conference, eds C. Preston, J. spread of the species into Canterbury, a region that Watts and N. Crossman, pp. 27-34 (Adelaide, is projected by the model to be optimally suitable Australia). climatically throughout its eastern districts, Kriticos, D.J. and Leriche, A. (2010). The effects of highlights the threat posed by the species and the spatial data precision on fitting and projecting utility of the model as a tool to guide its species niche models. Ecography In Press. management. Leathwick, J.R. and Stephens, R.T.T. (1998). Climate surfaces for New Zealand. Rep. No. ACKNOWLEDGMENTS LC9798/126. Landcare Research. We thank the Foundation for Research, Science and McLaren, D., Stajsic, V. and Iaconas, L. (2004). Technology, New Zealand, for funding this research The distribution, impacts and identification of (Undermining Weeds, C10X0811). exotic stipoid grasses in Australia. Plant Protection Quarterly 19, 59-66. REFERENCES McLaren, D.A., Stajsic, V. and Gardener, M.R. Anonymous (2003). Weed Management Guide - (1998). The distribution and impact of Chilean needle grass (Nassella neesiana). CRC South/North American stipoid grasses (Poaceae: for Weed Management. Stipeae) in Australia. Plant Protection Bell, M.D. (2006). Spread of Chilean needle grass Quarterly 13, 62-70. (Nassella neesiana) in Marlborough, New MDC (2009). Containment control pests. Zealand. New Zealand Plant Protection 59, http://www.marlborough.govt.nz/content/docs/e 266-270. nvironmental/regulatory/RPMS_07_P32-36.pdf. Bourdôt, G.W. and Hurrell, G.A. (1989). Ingress of 14 September. Stipa neesiana Trin. and Rupr. into swards of New, M., Hulme, M. and Jones, P. (1999). Lolium perenne L., Dactylis glomerata L. and Representing twentieth-century space-time Phalaris aquatica L., as affected by fertiliser climate variability. Part 1: Development of a and 2,2-DPA. New Zealand Journal of 1961-90 mean monthly terrestrial climatology. Agricultural Research 32, 317-326. Journal of Climate 12, 829-856. Bourdôt, G.W. and Ryde, D.H. (1986). Chilean Snell, K., Grech, C. and Jamie, D. (2007) 'National needle grass Stipa neesiana - Significance, Best Practice Management Manual - Chilean Idenification, Control. In Aglink, Vol. FPP 657, Needle Grass' (Victorian Government, pp. 1-2. Ministry of Agriculture and Fisheries, Melbourne). Wellington, New Zealand. Stace, C. (2001) 'New Flora of the British Isles', 2nd Conner, H.E., Edgar, E. and Bourdôt, G.W. (1993). edn. (Cambridge University Press, Cambridge). Ecology and distribution of naturalised species Sutherst, R.W., Maywald, G.F. and Kriticos, D.J. of Stipa in New Zealand. New Zealand Journal (2007). CLIMEX Version 3: User's Guide. of Agricultural Research 36, 301-307. Hearne Scientific Software Pty Ltd. Gardener, M.R., Whalley, R.D.B. and Sindel, B.M. (2003). Ecology of Nassella neesiana, Chilean needle grass, in pastures on the Northern Tablelands of New South Wales. I. Seed Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

24 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Part 3: Plants for possible inclusion in a RPMS

Environment Canterbury commissioned the following report A report on the plants to be considered for inclusion in the Canterbury Regional Pest Management Strategy Carol Jensen (20 Hyndhope Rd, Christchurch) April 2010.

3.1 Introduction Plants considered: 1. Moth plant (Araujia sericifera) 2. Smilax (Asparagus asparagoides) 3. Climbing asparagus (Asparagus scandens) 4. Rough horsetail (Equisetum hyemale) 5. Chilean Rhubarb (Gunnera tinctoria) 6. Senegal Tea (Gymnocoronis spilanthoides) 7. Giant hogweed (Heracleum mantegazzianum) 8. Yellow flag iris (Iris pseudacorus) 9. Purple loosestrife (Lythrum salicaria) 10. Yellow water lily (Nuphar lutea) 11. Chilean flamecreeper (Tropaeolum speciosum) 12. Green goddess (Zantedeshia spp.) 13. Bomarea (Bomarea caldasii) 14. Madeira vine (Anredera cordifolia) 15. False tamarisk (Myricaria germanica) 16. Royal fern (Osmunda regalis) 17. Asiatic knotweed (Reynoutria japonica) 18. Giant knotweed (Reynoutria sachalinensis) 19. African club moss (Selaginella kraussiana) 20. Grey willow (Salix cinerea) 21. Japanese spindle tree (Euonymus japonicus) 22. Pigs ear (Cotyledon orbiculata) 23. Cotoneaster simonsii 24. Puna grass (Achnatherum caudatum) 25. Russell lupin (Lupinus polyphyllus) 26. Boxthorn (Lycium ferocissimum) 27. Common polypody (Polypodium vulgare) 28. Carex pendula 29. Barberry (Berberis glaucocarpa) 30. Elm (Ulmus sp.) 31. Vipers bugloss (Echium vulgare) 32. (Muehlenbeckia australis) 33. Townsville stilo (Lotus sp.)

25 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

3.2 Background Section 72(1)(c) of the Biosecurity act 1993 states that: In order to justify the inclusion of the organism in a regional pest management strategy, a regional council must be of the opinion that: (c) The organism is capable of causing at some time a serious adverse and unintended effect in relation to the region on one or more of the following: (i) Economic wellbeing; or (ii) The viability of threatened species of organisms, the survival and distribution of indigenous plants or animals, or the sustainability of natural and developed ecosystems, ecological processes, and biological diversity; or (iii) Soil resources or water quality; or (iv) Human health or enjoyment of the recreational value of the natural environment; or (v) The relationship of Maori and their culture and traditions with their ancestral lands, waters, sites, waahi tapu, and taonga.

This report considers the extent of any adverse effects to the Canterbury region of each plant. The degree of risk posed by each plant is evaluated using the weed risk assessment method described in (Williams and Newfield, 2002). This system was developed taking the best features of existing systems (including Esler, 1993) and creating a comprehensive system using available information. Williams system was further refined, simplified and tested so that it could be applied to any region in NZ (Williams et al, 2005). The method is designed to enable ranking of new weeds in order of priority for control. Hence the scoring is weighted towards the most recent arrivals (those at an early stage of invasion) as these tend to be the easiest to control.

The weed risk assessment for each species considers and scores 4 attributes: 1) impact on vegetation and conservation values 2) invasion stage both in NZ and in the Canterbury region 3) biological success (ability to establish and persist) 4) public perception – difficulty in gaining acceptance for control

The resulting total score gives a ranking in order of priority for control. The scores and rank for each species, are listed in Appendix 1.

The species are considered in no particular order but are roughly grouped as follows: • Plants 1-12 are ranked high on the National Pest Plant Accord (NPPA) and adverse effects justify inclusion in the Regional Pest Management Strategy (RPMS). • Plants 13-23 are ranked low or medium on the NPPA but adverse effects justify inclusion in the RPMS. • Plants 24-29 are not on the NPPA but serious adverse effects justify inclusion in the RPMS. • Plants 30-32 adverse effects are not serious enough to justify inclusion in the RPMS. • Plant 33 not considered for inclusion on the RPMS.

26 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

3.3 Evaluation

1. Moth plant (Araujia sericifera)

Description Moth plant was introduced to NZ in the 1880s as an ornamental plant and has now become a serious pest in warmer areas of the country, being naturalised from Northland to Nelson/Marlborough. Moth plant is a rapid growing perennial climber. The vine can climb into the canopy where it can smother the vegetation below. The plant has very high seed production and seed viability, with seeds remaining viable for up to 5 years. The light fluffy seeds are wind dispersed over large areas and seedlings are shade tolerant. A milky sap is produced from the vine that can be a skin irritant and is toxic to humans and animals. In the past it has been promoted as an alternative food for monarch butterfly caterpillars, with people cultivating it for this purpose. Moth plant is difficult to control as vines can resprout from stumps and bared areas reseed profusely. Physical control is by pulling out seedlings, digging out vines and removing any seed pods. Chemical control is by cutting vines and coating stems with herbicide. Due to its rapid spread and ability to damage natural ecosystems some work has been done on investigating biological control. Moth plant is listed on the National Plant Pest Accord (NPPA) which means that it cannot be sold, propagated or distributed in New Zealand. Current and potential habitat invasion Although it is more common the warmer regions of the North Island it can invade lowland and coastal habitats and is increasingly being found in the warmer areas of the South Island including Marlborough. In Canterbury moth plant is regarded as ‘casual’ and is presently known only from gardens in Christchurch and on the Plains (Mahon, 2007). It is not known to have escaped into the wild. It has a very wide environmental tolerance to drought, humidity, wind, salt and a range of soils. It was thought to be limited by frost but it has shown that it can withstand the colder conditions in Christchurch gardens. Moth plant is rapidly spreading in warm and coastal northern regions and the northern South Island and it is considered that this invasive vine has a much wider potential distribution. Should it escape from its current limited distribution in Canterbury moth plant has the potential to invade shrubland and forest on coastal and lowland habitats and on Banks Peninsula. Risks [under Section 72(1)(c)] Moth plant is considered to have economic, conservation and health impacts. • Economic impacts Currently moth plant has ‘casual’ status in Canterbury. Should it become fully naturalised then moth plant is likely to have costly economic impacts due to the cost of control. Moth plant is a NPPA species so where it grows near nursery outlets, abundant wind-born seed may contaminate pot plants so creating issues for nursery inspections and enforcement of the NPPA. • Conservation impacts Moth plant is a fast growing shade-tolerant vine that rapidly smothers and replaces native vegetation, invading intact or disturbed forest, forest margins, shrublands and open habitats e.g. coastal areas. It can cause physical harm to trees by strangling the stems and the weight of the vine breaking branches. The flowers can kill insects, butterflies and moths by trapping them within the flower. Although moth plant is more common in warmer North Island areas it has the potential to invade Canterbury shrublands, forests and natural areas. • Health impacts The moth plant has a milky sap that is toxic to animals and humans. The pods can cause a severe reaction if swallowed and the sap can be irritating to the skin.

27 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Section 72(1)(c) Using the weed risk assessment method of Williams (2005) moth plant is assessed as being capable of causing serious adverse effects should moth plant become naturalised in Canterbury. The score reflects that moth plant is at an early stage of invasion in Canterbury. Spread into the wild will be less likely if control is planned at this early stage of invasion. The weed risk assessment of moth plant in the NPPA also ranks the weed risk as high priority. The potential adverse effects as described here provide justification for the inclusion of moth plant in the RPMS (Section 72(1)(c)).

28 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Moth plant (Araujia sericifera)

Points Score A. Interactions 1 Volume of individual plant m3: 10, 100, 1000, 10 000 1 to 4 3 2 Totally pre-empts sites, or covers native species to form canopy 2 or 0 2 3 Growth appears faster than associated native species 1 or 0 1 4 Species persists: < 5 years, 5–20 years, > 20 years 1 to 3 3 Impact score (Sum A1–4) 9

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 0 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 4 numerous small (2), numerous large (0)

D. Cultivation and perceptions 1 Present as: mass plantings (3), frequent smaller plantings (2), 3 to 0 1 infrequent small plantings (1), not planted (0) 2 No. nurseries selling species: > 3 ,< 3, 3 to 0 0 3 Is it a crop plant? 1 or 0 0 4 Does it have unpleasant features? 1 or 0 1 Public attitudes score (sum D1–3 minus D4) 0

29 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

2. Smilax (Asparagus asparagoides)

Description Smilax has tough wiry stems and a smothering growth habit. It forms a canopy over plants 2- 3m high, even in shade. It is a serious weed in Australia, where it is known as bridal creeper. In the past it was often grown as an ornamental but has now escaped into the wild. It is now included on the NPPA. It invades disturbed forest and margins, coastal areas and roadsides. Smilax prefers fertile, well-drained, lightly-textured soils but tolerates all but the wettest soils. Smilax has small sticky red berries containing tiny black seeds. Dispersal is via birds, animals and machinery and tubers can sprout where garden rubbish is dumped. Control of smilax is difficult due to the long-lived tubers that can resprout. It can be controlled with chemicals but it is often difficult to detect outlying plants before they have fruited. Current and potential habitat invasion Smilax is widely established but scattered in the North Island and is present in the Nelson Marlborough region. It is regarded as fully naturalised in Canterbury (Mahon, 2007) but is currently rare and local in the wild around Christchurch and Banks Peninsula. However it is occasionally present in gardens in Christchurch, Banks Peninsula and on the Plains. Smilax is a very troublesome weed with the potential for much further spread in New Zealand. In Canterbury smilax has the potential to invade shrublands and regenerating forest on Banks Peninsula, coastal areas and the foothills. Risks [under Section 72(1)(c)] Smilax is considered to have economic, conservation and health impacts.

• Economic impacts Should Smilax become more widespread then it is likely to have costly economic impacts due to the cost of control. • Conservation impacts The wide environmental tolerances, prolific seed production and smothering growth habit of smilax means that it has the ability to invade shrubland and forest on Banks Peninsula, coastal areas and inland foothills. The vine is capable of smothering and dominating scrub, open clearings and the forest understorey. It is a serious threat to native plant communities as it can out-compete other vegetation by forming pure colonies. Although smilax is currently rare and local in its distribution the impacts would be much greater if it became more widespread. • Health impacts There are numerous reports of allergy to Asparagus plants in the medical literature. The berries and uncooked shoots in particular can be toxic. Contact dermatitis is often described. Section 72(1)(c) The high score from the weed risk assessment reflects the fact that smilax is a serious weed but is at an early stage of invasion in Canterbury. Early intervention at this stage should limit the extent of smilax and prevent it from becoming established in new areas. The potential adverse effects, as described here, provide justification for the inclusion of smilax in the RPMS (Section 72(c)).

30 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Smilax (Asparagus asparagoides) Points Score A. Interactions 1 Volume of individual plant m3: 10, 100, 1000, 10 000 1 to 4 1 2 Totally pre-empts sites, or covers native species to form canopy 2 or 0 2 3 Growth appears faster than associated native species 1 or 0 1 4 Species persists: < 5 years, 5–20 years, > 20 years 1 to 3 3 Impact score (Sum A1–4) 7

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 4 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 4 numerous small (2), numerous large (0)

D. Cultivation and perceptions 1 Present as: mass plantings (3), frequent smaller plantings (2), 3 to 0 0 infrequent small plantings (1), not planted (0) 2 No. nurseries selling species: > 3 ,< 3, 3 to 0 0 3 Is it a crop plant? 1 or 0 0 4 Does it have unpleasant features? 1 or 0 0 Public attitudes score (sum D1–3 minus D4) 0

31 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

3. Climbing asparagus (Asparagus scandens)

Description Climbing asparagus is a native of South Africa and was introduced to NZ as an ornamental garden plant. It has now escaped into the wild although it is still a popular plant because of its feathery foliage. It is now included on the NPPA list. Climbing asparagus is a scrambling vine that is capable of smothering native vegetation. It is shade tolerant so can scramble over forest floor vegetation up to a height of 4 metres. Thus it can prevent regeneration of native species and can strangle soft barked trees and shrubs by wrapping itself around trunks. The plant has the potential to be a major weed in forest margins, undisturbed forests, native forest remnants, natural open areas, roadsides and riverbanks. Climbing asparagus is spread mostly by birds eating the berries and distributing the seeds in their droppings. It is also spread by the dumping of garden waste. Once established it spreads quickly by root tubers and running stems. Current and potential habitat invasion Infestations of climbing asparagus are found throughout the country but it is not common in the South Island. There are scattered infestations in Marlborough but in Canterbury it is only known from several garden locations in Christchurch and Banks Peninsula. There are 2 records under second growth trees on Banks Peninsula (Mahon, 2007). It is regarded as fully naturalised in New Zealand but only ‘casual’ in Canterbury. Climbing asparagus is considered a very troublesome weed with potential to spread further. Although presently limited in its distribution it has the potential to invade large areas of shrublands, regenerating forest and open areas on Banks Peninsula and throughout Canterbury. Risks [under Section 72(1)(c)] Climbing asparagus is likely to have some economic impacts and a big impact on conservation values should it become fully naturalised in Canterbury. It also has health impacts which apply to asparagus plants in general. • Economic impacts Should Smilax become more widespread then it is likely to have costly economic impacts due to the cost of control. • Conservation impacts The wide environmental tolerances, prolific seed production and smothering growth habit of climbing asparagus means that it has the ability to invade shrubland, forest (particularly the understorey), open areas, roadsides and riverbanks on Banks Peninsula and throughout Canterbury. Although climbing asparagus is currently rare and local in its distribution the impacts would be much greater should this weed become more widespread and establish in new areas. • Health impacts There are numerous reports of allergy to Asparagus plants in the medical literature. The berries and uncooked shoots in particular can be toxic. Contact dermatitis is often described. Section 72(1)(c) The high score from the weed risk assessment reflects the fact that climbing asparagus is a serious weed but is at an early stage of invasion. Its current limited distribution in Canterbury means that will be cheaper and need less effort to control than more widespread and established weeds. The potential serious adverse effects as described here provide justification for the inclusion of climbing asparagus in the RPMS (Section 72(1)(c)).

32 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Climbing asparagus (Asparagus scandens)

Points Score A. Interactions 1 Volume of individual plant m3: 10, 100, 1000, 10 000 1 to 4 2 2 Totally pre-empts sites, or covers native species to form canopy 2 or 0 2 3 Growth appears faster than associated native species 1 or 0 1 4 Species persists: < 5 years, 5–20 years, > 20 years 1 to 3 3 Impact score (Sum A1–4) 8

C. Reproduction 1 Species cryptic and cannot be detected before it reproduces 1 or 0 1 2 Produces viable seed 2 or 0 2 3 Seed dispersed primarily by: small birds, wind, or water (2), 2 or 1 2 large birds or passive/accidental dispersal (1) 4 Minimum generation time < 3 years (2), > 3 years (1) 2 or 1 1 5 Persistent vegetative organs above or below ground, 2 or 0 2 or seed bank (> 1 year) 6 Juveniles common within 100 m parents 1 or 0 1 Spread score (sum B1–3 + C1–6) 17 Impacts x spread score 136 D. Cultivation and perceptions 1 Present as: mass plantings (3), frequent smaller plantings (2), 3 to 0 1 infrequent small plantings (1), not planted (0) 2 No. nurseries selling species: > 3 ,< 3, 3 to 0 0 3 Is it a crop plant? 1 or 0 0 4 Does it have unpleasant features? 1 or 0 1 Public attitudes score (sum D1–3 minus D4) 0

33 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

4. Rough horsetail (Equisetum hyemale)

Description Rough horsetail is a perennial that grows up to 2 metres tall. It has extensive underground rhizomes and ridged hollow stems. Spores are produced in cone-like structures on fertile stems. It is sometimes grown as an ornamental plant due to its unusual appearance. An attractive plant it is sometimes grown in pots and used in floral arrangements. Horsetail is also used in natural medicine. It is most commonly spread by humans selling or giving it away as a medicinal herb.

Rough horsetail prefers moist areas such as gravel and pond/lake margins but once it is well established, it will adapt to a wide range of conditions. It has potential to form tall dense masses in wetland and damp open places. It spreads rapidly by underground rhizomes and it is extremely difficult to control once established. It can displace desirable plant species and is usually spread via the movement of soil containing rhizomes, or through deliberate planting. All Equisetum spp. are difficult to control with chemicals and almost impossible to remove manually. Rough horsetail is classed as an unwanted organism and is listed in the NPPA. Current and potential habitat invasion Another Equisetum species, field horsetail (Equisetum arvense), is fully naturalized in Canterbury but rough horsetail is classified as ‘casual’ (Mahon, 2007). It was first recorded in the wild in Christchurch in 1994 (Webb et al. 1995). Most known sites in Canterbury are in gardens. An aggressive plant that is almost impossible to eradicate, it has potential to form tall dense masses in wetland and damp open places so could potentially clog waterways throughout Canterbury. Risks [under Section 72(1)(c)] Rough horsetail has potential economic, conservation and health impacts. • Economic impacts Should rough horsetail become more widespread then it is likely to have costly economic impacts due to the cost of control. • Conservation impacts Rough horsetail has the potential to cause serious damage to wetland ecosystems were it to become established in the wild in Canterbury. Wetlands contain a wide variety of native plants, fish (including the rare Canterbury mudfish) and invertebrate species. It is also likely to compete with a range of native plants including flax and raupo. • Health impacts The plant is toxic to stock. Equisetum spp. are rich in thiaminase, an enzyme that destroys thiamine, an essential amino acid. Monogastric animals like horses are poisoned, the animals becoming increasingly unthrifty and have difficulty breathing. These plants also contain nicotine (an alkaloid), so the entire plant is toxic with the roots and stem base are most toxic. Although it can be used as a medicinal herb the plants are toxic to humans if consumed in large quantities. Section 72(1)(c) The high score from the weed risk assessment reflects the fact that rough horsetail is a serious weed but is at an early stage of invasion in Canterbury. Early intervention at this stage should prevent rough horsetail from becoming established in new areas.

The potential serious adverse effects, as described here, provide justification for the inclusion of rough horsetail in the RPMS (Section 72(1)(c)).

34 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Horsetail (Equisetum hyemale)

Points Score A. Interactions 1 Volume of individual plant m3: 10, 100, 1000, 10 000 1 to 4 1 2 Totally pre-empts sites, or covers native species to form canopy 2 or 0 2 3 Growth appears faster than associated native species 1 or 0 1 4 Species persists: < 5 years, 5–20 years, > 20 years 1 to 3 3 Impact score (Sum A1–4) 7

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 8 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 4 numerous small (2), numerous large (0)

35 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

5. Chilean Rhubarb (Gunnera tinctoria)

Description Chilean rhubarb is a giant clump-forming plant with large rhubarb, umbrella-shaped leaves up to 2.5 m high. Plants die back in winter to exposed clumps of roots. It is a popular garden plant and is widely grown as a waterside plant in parks, botanic gardens and large private gardens but is now on the NPPA list.

Chilean rhubarb prefers to grow in moist places and will invade forests, waterways, roadsides, swamp margins, slips and wet coastal cliffs. It is a big problem on the coastal cliffs of Taranaki. Natural dispersal to new sites appears to be rather slow except where suitable habitat is continuous, such as the Taranaki cliffs. Chilean rhubarb can also block and restrict access to waterways. It is a shade-tolerant plant and it can be spread by seed via birds or water. It is also spread by stem fragments.

Adult plants and seedlings are relatively easy to control with appropriate sprays. However spraying this weed on cliffs or other inaccessible habitats may be difficult.

Current and potential habitat invasion Chilean rhubarb is widespread in western Taranaki and Wanganui, scattered elsewhere in the North Island and scattered in the South Island, particularly on the West Coast. It has recently been recorded on Stewart Island where it is considered to pose a significant threat to indigenous biodiversity (Heenan et al. 2009). Although of scattered occurrence it has a very wide geographical range from Stewart Island to Northland and is limited mainly by suitable habitat of moist colluviums.

In Canterbury it is fully naturalised (Mahon, 2007) although it is most common in gardens and around ponds. Wilson (1999) noted some Chilean rhubarb established on roadside banks at Okains Bay, Banks Peninsula and the Department of Conservation currently controls small populations on Banks Peninsula.

Potential habitat in Canterbury includes all wetland areas, river banks, roadsides and forest margins. Although it is common on Taranaki coastal cliffs, the coastal cliffs of Canterbury are probably too dry for Chilean rhubarb to establish.

Risks [under Section 72(1)(c)] Chilean rhubarb is considered to have economic and conservation impacts. • Economic impacts Should Chilean rhubarb become more widespread then it is likely to have costly economic impacts due to the cost of control. Chilean rhubarb can also block and restrict access to waterways resulting in the increased cost of clearing the waterways should the weed become established. • Conservation impacts Chilean rhubarb smothers native communities and may completely transform the landscape. It grows quickly, forming large clumps that shade out native plant species beneath it.

Section 72(1)(c) The high score from the weed risk assessment reflects the fact that Chilean rhubarb is a serious weed but is at an early stage of invasion. It is likely to be easier to control (and therefore cheaper) than other more widespread and established weeds. Chilean rhubarb

36 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

should be considered for eradication in Canterbury as the small populations at a limited number of sites means that eradication is realistic and feasible.

The potential adverse effects as described here provide justification for the inclusion of Chilean rhubarb in the RPMS (Section 72(c)).

Weed Risk Assessment

Chilean rhubarb (Gunnera tinctoria)

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 6 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 4 numerous small (2), numerous large (0)

D. Cultivation and perceptions 1 Present as: mass plantings (3), frequent smaller plantings (2), 3 to 0 2 infrequent small plantings (1), not planted (0) 2 No. nurseries selling species: > 3 ,< 3, 3 to 0 0 3 Is it a crop plant? 1 or 0 0 4 Does it have unpleasant features? 1 or 0 0 Public attitudes score (sum D1–3 minus D4) 2

37 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

6. Senegal Tea (Gymnocoronis spilanthoides)

Description Senegal tea is a perennial aquatic herb which can grow up to 1m high. The plant is dormant over winter when it dies back to the rootstock then resprouts in spring. It grows in wetland communities in wet marshy soils, at water margins and in still or flowing water. It is an attractive fast growing ornamental pond plant with scented white flowers attractive to butterflies. It is also used in aquariums as a submerged plant.

Senegal tea produces viable seed. Vegetative reproduction occurs through the production of roots at stem nodes and vegetative fragmentation. Dispersal is by seed stems and root fragments being dispersed in water, on livestock hooves and machinery. It can also be spread by dumped aquaria contents when liberating fish.

The plant is tolerant of shade, frost and poor drainage but intolerant to drought. It is highly invasive and is now included on the NPPA. It is difficult to control, both through lack of effective methods and its aquatic habitat.

Current and potential habitat invasion There are scattered garden sites in the North Island and in Marlborough. In Canterbury the only known sites are on either side of the Waimakariri River where it has probably been dumped with garden rubbish.

There is high potential for Senegal tea to spread in rivers and wetlands, in particular where rubbish is likely to be dumped near centres of population (eg. Avon, Heathcote, Halswell and Waimakariri Rivers). Should Senegal tea become established it is likely that most waterways and wetlands in Canterbury would be at risk.

Risks [under Section 72(1)(c)] Senegal tea is likely to have economic and conservation impacts. • Economic impacts Senegal tea poses potential negative economic impacts in Canterbury due to the potential cost of control and the need for additional drain and waterway clearance should the weed become established. • Conservation impacts Senegal tea is invasive in fertile wetlands, flowing and still waters, impacting on biodiversity values and promoting flooding to the detriment of the native aquatic flora and fauna. It grows quickly forming dense floating mats which exclude all other plants and impacts on native fish and invertebrates.

Section 72(1)(c) The high score from the weed risk assessment reflects the fact that Senegal tea has potential to be a serious weed if it becomes naturalised in Canterbury. Its current known distribution is localised and limited so it could be eradicated. The potential adverse effects as described here provide justification for the inclusion of Senegal tea in the RPMS (Section 72(1)(c)).

38 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Senegal tea (Gymnocoronis spilanthoides)

39 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

7. Giant hogweed (Heracleum mantegazzianum)

Description Giant hogweed is a large biennial herb which can grow up to 3m tall. It has large serrated leaves and produces many long lived flattened seeds. Plants get frosted and die back to the basal root clump in winter. It is sometimes cultivated as an ornamental curiosity. It is fully naturalized in New Zealand and is now listed on the NPPA. It is often found in cold damp places in gardens, waste places and riparian areas. Seeds appear to germinate well if there is a cold damp spring.

The plant exudes a clear watery sap which sensitizes the skin to ultraviolet radiation. This can result in severe burns to the affected areas resulting in severe blistering and painful dermatitis.

Giant hogweed can also compete with and exclude native vegetation that grows along river or stream edges. The above ground parts are easy to knock down with glyphosate, but persistent rootstocks can be hard to kill.

Current and potential habitat invasion Giant hogweed is found scattered throughout Canterbury. It can be found growing in gardens, waste places and along river/stream sides.

Giant hogweed has the potential to spread along riparian areas and around ponds and lakes.

Risks [under Section 72(1)(c)] Giant hogweed is considered to have conservation and health impacts. • Conservation impacts Giant hogweed can have a serious impact on conservation values especially along river and stream banks where it can form a dense canopy out-competing native riparian species, resulting in increased soil erosion along stream banks. This could be serious where riparian conservation sites may be fenced off from stock. • Health impacts Giant hogweed is poisonous. The plant exudes a watery, clear sap, which on direct contact with skin can cause sensitivity to ultra violet radiation. This can result in painful burns and blisters. Exposure to even small particles of giant hogweed sap or dust exposed can irritate the skin. The bristles on the stalks and stem also contain a toxic sap.

Section 72(1)(c) The high score from the weed risk assessment reflects the fact that giant hogweed has weedy characteristics but is presently mainly confined to gardens and waste places. The potential adverse effects as described here provide justification for the inclusion of giant hogweed in the RPMS (Section 72(1)(c)).

40 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Giant hogweed (Heracleum mantegazzianum)

41 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

8. Yellow flag iris (Iris pseudacorus)

Description Yellow flag is a robust aquatic perennial to 1-2 m that grows in leafy clumps and forms dense rhizomes. It has tall stems and yellow flowers.

It is of concern as a weed because of the rhizomes which form dense floating mats that may overtop native species that grow on margins of water bodies. It is tolerant of saline, frost, flooding and drought, high-low fertility, many soil types and damage to stems. It is poisonous, so is usually not grazed by stock.

Seeds and rhizome fragments are spread by water and contaminated machinery. It is a ‘garden escape’ plant that has spread from gardens and deliberate plantings into the environment. Yellow flag is on the NPPA list which should reduce the spread to new catchments.

The Christchurch City Council has been working on control of some large areas of yellow flag in the lower Avon River. A combination of cutting and removal and injecting the rhizomes with herbicide are proving effective in the gradual reduction of the infestation.

Current and potential habitat invasion Yellow flag is scattered throughout New Zealand and may be locally common. In Canterbury the lower Avon River contains one of the largest populations in New Zealand and pockets occur in the Halswell River. It is also common in gardens.

It has the potential, if unchecked, to spread into wetlands (including salt marsh), estuaries, wet pasture and river margins throughout lowland Canterbury. In particular, important wetlands including Travis Swamp and Te Waihora Lake Ellesmere are at risk.

Risks [under Section 72(1)(c)] Yellow flag is considered to have economic, conservation and health impacts. • Economic impacts Yellow flag is capable of blocking drains and waterways resulting in the increased cost of clearing dense mats from drains, rivers and wetlands should the weed become established. • Conservation impacts Yellow flag invades swampy ground, fresh or brackish water margins, lakes, salt marsh, and wet sandy areas. Rhizome mats can displace native plants, especially vulnerable species that live on the margins of water bodies. The mats can cause flooding and changes in water level in swamps and may compete with native species eg. flax and raupo. Poisonous seeds may have an impact on birdlife. • Health impacts This species can cause contact dermatitis and allergies in some people, via contact with sap or seeds in particular. Many plants in this genus are poisonous if ingested, with roots and leaves toxic to animals, including humans. Iris species have poisoned cattle and swine and may cause similar symptoms in humans if the rhizomes are ingested (Connor, 1977).

Section 72(1)(c) The medium score from the weed risk assessment reflects the fact that yellow flag has potential to be a serious weed if it becomes more widespread. Its current distribution is localised and limited so could be controlled relatively easily. The potential serious adverse

42 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

effects as described here provide justification for the inclusion of yellow flag in the RPMS (Section 72(c)).

Weed Risk Assessment

Yellow flag iris (Iris pseudacorus)

43 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

9. Purple loosestrife (Lythrum salicaria)

Description Purple loosestrife is an erect summer-green perennial herb which can grow up to 3m tall. It has a taproot and fibrous roots that form dense mats that produce many stems of showy purple – magenta flowers followed by numerous long-lived and viable seeds. The above ground parts die down to the basal root stock in winter. Purple loosestrife is a garden escape that was widely cultivated but is now listed on the NPPA which should reduce long distance spread.

It can quickly invade damp ground and shallow water and tolerates a wide range of temperature and nutrients but is intolerant of saline conditions. The dense bushy growth habit of purple loose-strife forms massive tall impenetrable stands that can overtop and exclude native species. It destroys wetland and marginal habitat and food sources for many fish and bird species. The dense stands can also impede flow and cause blockages resulting in flooding.

Spraying with glyphosate is effective but follow-up is needed as it re-sprouts profusely and seeds are long-lived.

Current and potential habitat invasion Purple loosestrife invades wetlands, lakesides, streams, damp and ephemeral bogs and can creep onto dry margins. It is widespread throughout New Zealand. In Canterbury it is widespread on lowland sites from Banks Peninsula to the foothills and from North to South Canterbury. Purple loosestrife populations have been declining in Canterbury due to control at known sites by Environment Canterbury, Christchurch City Council and the Department of Conservation. It has been dug out of many garden sites.

Should control be relaxed in Canterbury then there is potential for purple loosestrife to invade and reinvade many wetland sites throughout Canterbury including rivers, streams, drains, swamps and even dry waste sites.

Risks [under Section 72(1)(c)] Purple loosestrife is considered to have economic, conservation and recreational impacts. • Economic impacts Tall impenetrable stands of purple loosestrife may cause disruption to irrigation and drainage operations. There may be increased costs to clear drains and waterways if they become blocked. • Conservation impacts Dense stands of purple loosestrife may displace other marginal and wetland vegetation. Native vegetation may be excluded by being overtopped by dense stands of purple loosestrife. Oxygen depletion may destroy food sources and habitat of many fish and birds. • Recreational This plant may form very thick beds making boat traffic difficult and impeding access to water bodies. Death of fish may also occur in large numbers due to oxygen depletion so reducing the recreational value of fishing.

Section 72(1)(c) The high score from the weed risk assessment reflects the fact that purple loosestrife has potential to be a serious weed if it is not controlled. Its current distribution in Canterbury is widespread but is reducing due to current control programmes. In addition the weed risk assessment for the NPPA ranks the weed risk as high priority. The potential serious adverse

44 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

effects as described here provide justification for the inclusion of purple loosestrife in the RPMS (Section 72(c)).

Weed Risk Assessment

Purple loosestrife (Lythrum salicaria)

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 5 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 2 numerous small (2), numerous large (0)

45 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

10. Yellow water lily (Nuphar lutea)

Description The yellow water lily is an aquatic herb with large floating leaves and single yellow flowers. The root system is made up of large rhizomes which can be up to 3m deep. It is found in slow-running waterways and in lakes. In New Zealand it is only known from Hawkes Bay where it is restricted to one lake and adjacent dam. At this one site this species has displaced all other aquatic species from the shoreline to 2 m deep. Mahon (2007) does not record the presence of yellow water lily in Canterbury but there is an unconfirmed report of a stream choked with yellow water lily near Fairly in Canterbury (Ecan, 2005).

It is difficult to control, both through lack of effective methods and its aquatic habitat. However, it has been nearly eradicated at the one recorded site in Hawkes Bay.

It can only be spread by deliberate human planting and subsequent movement by water. It is included on the NPPA list which should prevent spread to new catchments.

Current and potential habitat invasion Yellow water lily is known from one site in Hawkes Bay and an unconfirmed report from near Fairly in Canterbury. However it could be potentially be problematic in most nutrient-rich water bodies in lowland New Zealand.

Should yellow water lily become established in Canterbury then all lowland waterways and lakes in Canterbury would be at risk of invasion. It is recommended that the unconfirmed report from Fairly is followed up. If this infestation is confirmed as yellow water lily then prompt action may prevent the spread of this potentially serious weed in Canterbury.

Risks [under Section 72(1)(c)] Yellow water lily is likely to have economic and conservation impacts should it become naturalised in Canterbury. • Economic impacts Yellow water lily is likely to clog drains and waterways so there would costly economic impacts associated with cost of control and mechanical clearance of the weed. • Conservation impacts As yellow water lily has shown it can exclude all other vegetation from the vicinity in which it grows the loss to native biodiversity would be of great concern. The destruction of habitat and food sources for invertebrates, fish and birds is also likely if this weed becomes naturalised.

Section 72(1)(c) The high score from the weed risk assessment reflects the fact that yellow water lily has potential to be a serious weed if it ever became established in Canterbury. The weed risk assessment for the NPPA also ranks the weed risk as high priority for this species. The potential serious adverse effects, as described here, provide justification for the inclusion of yellow water lily in the RPMS (Section 72(c)).

46 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk assessment

Yellow water lily (Nuphar lutea)

47 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

11. Chilean flamecreeper (Tropaeolum speciosum)

Description Chilean flamecreeper is a climbing perennial vine with attractive red flowers. It is in the same family as nasturtiums and is a garden escape. It is found mainly in remnant stands of forest, scrub and sometimes in more remote forest clearings. It can climb high into the canopy and its fruit is dispersed by birds. It is tolerant of shade, warm-cold temperatures, salt, wind, many soil types and damp to dry conditions.

It is difficult to control as it keeps regrowing from the roots after herbicide control or cutting.

Current and potential habitat invasion Chilean flamecreeper is widespread in New Zealand but only common in limited areas. It is more likely to invade cooler areas and is currently invasive in Southland, Canterbury and the Central North Island. It is also recorded on Stewart Island. In Canterbury it is present in higher rainfall bush reserve areas behind Oxford, in the Ashley Gorge, Waimakariri (Lords Bush), Peel Forest, Talbot Forest, Mt Nimrod and on Banks Peninsula.

Chilean flame-creeper is a serious threat to forest and shrubland especially in the higher rainfall bush areas of the Canterbury foothills and on Banks Peninsula.

Risks [under Section 72(1)(c)] Chilean flame-creeper is considered to have economic and conservation impacts. • Economic Chilean flamecreeper is difficult to control so there are economic impacts of ongoing control. • Conservation Chilean flamecreeper is a smothering vine that invades light gaps and forest edges, inhibits regeneration and competes with native plants. It can survive in the low light conditions found beneath the forest canopy and as it is spread by birds it is easily spread from one bush area to another.

Section 72(1)(c) The high score from the weed risk assessment reflects the fact that Chilean flamecreeper has potential to be a serious weed in Canterbury. The weed risk assessment for the NPPA also ranks the weed risk as high priority. The potential serious adverse effects, as described here, provide justification for the inclusion of Chilean flamecreeper in the RPMS (Section 72(c)).

48 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed risk assessment

Chilean flamecreeper (Tropaeolum speciosum)

49 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

12. Green goddess (Zantedeshia spp.)

Description Zantedeschia aethiopica (L.) Spreng. cv. Green Goddess is very similar to Zantedeschia aethiopica but has distinctive green flowers. The Zantedeschia species (arum lilies) originates from southern Africa and have been imported into New Zealand since the early 1900's. From these original imports, a range of hybrids have been produced over the years and this has formed the basis of a $10 million dollar export industry for cut flowers and bulbs. Arum lilies are popular with home gardeners as well as commercial growers (ERMA newsletter, 8 September 2000).

Arum lilies are evergreen, clump-forming tuberous perennial herbs which grow to 1.5m. In green goddess the amount of green on the flower can vary and in garden situations the white arum lily is known to sometimes revert to the green leaved green goddess form. It is likely that both are the same species with the Green goddess variety showing wide variation in leaf colour.

Arum lilies are found in swamps and open damp areas with low cover. They tolerate wet (drought-resistant once established), wind, salt, hot to cold, most soil types and moderate shade. Green goddess appears to have wider tolerances than the white arum lily being able to tolerate deep shade as well as full sunlight. Green goddess tolerates a wide range of habitats from brackish wetlands to flowing water and sand dunes. Green goddess seeds prolifically and the seeds are dispersed by birds and water movement. It is also spread by tubers, garden waste being an important source of new infestations. Clumps expand slowly via new shoots. Green goddess is a common garden escape forming colonies in the wild.

Current and potential habitat invasion A very popular ornamental species, Zantedeschia aethiopica is cited as invasive in many countries, from hot and moist temperate zones to tropical and subtropical zones. It grows in dense stands preventing the regeneration of local flora. In Canterbury arum lily is regarded as fully naturalised but it is not known how much of this is the Green goddess variety. The distribution of green goddess appears very local with few known large infestations but probably many small garden escapes. It is present near Gore Bay and there are small pockets in Cashmere Stream and the Halswell River in Christchurch. At Barrys Bay on Banks Peninsula there is a small infestation in a muddy creek but it does not appear to be expanding (Miles Giller, pers.comm).

In Canterbury there is potential for green goddess to invade swampy areas and damp wasteland.

Risks [under Section 72(1)(c)] Green goddess is considered to have economic, conservation and health impacts. • Economic impacts Green goddess provides both positive and negative economic impacts. A positive impact arises from both the white arum lily and green goddess being grown commercially for floral arrangements.

Potential negative impacts arise from both the white and green arum lilies invading wet pasture so could be regarded as agricultural weeds. They are unpalatable and therefore have a competitive advantage in heavily grazed situations. • Conservation impacts

50 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Green goddess is a persistent colony-forming invader of swampy areas, smothering the ground and preventing the establishment of native seedlings. It is spread by birds and is particularly invasive in wet areas where it may compete with native plants eg. flax and raupo. • Health impacts Zantedeschia aethiopica is also highly toxic and can cause death if ingested by humans or livestock.

Section 72(1)(c) The high score from the weed risk assessment reflects the fact that Green goddess has potential to be a serious weed if it becomes widely established. Its current distribution is localised and limited but there are potential economic, conservation and health impacts. The potential serious adverse effects as described here provide justification for the inclusion of Green goddess in the RPMS (Section 72(c)).

51 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Green goddess (Zantedeshia spp.)

D. Cultivation and perceptions 1 Present as: mass plantings (3), frequent smaller plantings (2), 3 to 0 1 infrequent small plantings (1), not planted (0) 2 No. nurseries selling species: > 3 ,< 3, 3 to 0 0 3 Is it a crop plant? 1 or 0 1 4 Does it have unpleasant features? 1 or 0 1 Public attitudes score (sum D1–3 minus D4) 1

52 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

13. Bomarea (Bomarea caldasii)

Description Bomarea is a multi-stemmed vine of the lily family that twines around any available support. It has reddish flowers and large bright orange fruit. The root system is made up of rhizomes and numerous tubers. Bomarea is a fast growing plant that climbs up into the canopy where it forms a huge mass that that smothers and can kill the trees supporting it. It also obstructs light reaching the ground which prevents native seedlings from establishing. Birds can disperse the fruit large distances to inaccessible places, which can make control difficult.

Control is by glyphosate or vigilant gel to cut stems.

The attractive flowers of Bomarea make it an attractive garden plant. It is now listed on the NPPA which should limit it spreading to new areas. Bomarea invades forest margins and disturbed forest remnants.

Current and potential habitat invasion Flora III (Healy and Edgar, 1980) lists only one occurrence of Bomarea as a garden escape in Auckland. Herbarium records since then indicate it is now more common and widespread. Although only naturalised for a short period it is becoming more common and has the potential to cause serious impacts.

Bomarea is a problem in bush remnants in Dunedin and on the Otago Peninsula. It has recently been recorded around old house sites on Stewart Island (Heenan et al. 2009). In Canterbury it is only known from gardens and nearby hedges and bush edges in Christchurch, Governors Bay and Timaru (Mahon, 2007).

Should it escape from gardens then regenerating bush and native forest will be at risk of invasion especially on Banks Peninsula.

Risks [under Section 72(1)(c)] Bomarea is considered to have economic, conservation and health impacts. • Economic impacts Should Bomarea become more widespread then it is likely to have economic impacts due to the cost of control. • Conservation impacts Bomarea is a serious threat to bush remnants and regenerating shrubland and forest. It forms a dense cover thereby smothering the plants beneath and preventing seedling establishment. Although its distribution in Canterbury is mostly near gardens its impacts would be much greater should it become more widespread. • Health impacts Bomarea spp. appear to be a natural source of the compound alpha-methylene- gamma-butyrolactone, which causes plant contact dermatitis.

Section 72(1)(c) The high score from the weed risk assessment reflects the fact that Bomarea is a serious weed but is at an early stage of invasion. Bomarea should be considered for eradication in Canterbury as the small populations at a limited number of sites means that eradication is realistic and feasible.

The potential serious adverse effects, as described here, provide justification for the inclusion of Bomarea in the RPMS (Section 72(1)(c).

53 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Bomarea (Bomarea caldasii)

54 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

14. Madeira vine (Anredera cordifolia)

Description Madeira vine is a woody perennial climbing vine with fleshy rhizomes and reddish stems that produce small aerial tubers. Although many white fragrant flowers are produced it does not produce seed, hence the main means of dispersal is by the aerial tubers breaking off the stems. The tubers are known to survive up to 5 years and are very difficult to kill. Hand removal is difficult as the tubers break off easily. Although herbicide kills the foliage the tubers are more resistant. Dispersal to new areas is mainly by the dumping of garden rubbish resulting in the tubers, stems or roots resprouting.

It is not much planted by gardeners now but it was once considered a desirable ornamental plant. This species is still quite restricted and requires human help to disperse, hence, it is listed on the NPPA.

Madeira vine is most commonly found in coastal areas and scrub covered gullies especially in warmer areas of the North Island. It is tolerant of drought, damp, wind, salt and shade.

Current and potential habitat invasion Madeira vine is widespread throughout New Zealand although not particularly common. It is common in Auckland and Northland but at present uncommon in Wellington, Nelson, Marlborough and Canterbury. Although Madeira vine is regarded as fully naturalized in Canterbury it is known from only a few roadside sites around Christchurch.

Should Madeira vine escape from its current rare and local distribution in Canterbury it has the potential to invade shrubland and forest margins on coastal and lowland habitats and on Banks Peninsula.

Risks [under Section 72(1)(c)] Madeira vine is considered to have economic and conservation impacts. • Economic impacts Should Madeira vine become more widespread it is likely to have an economic impact due to the cost of control. • Conservation impacts Madeira vine invades regenerating shrubland and forest margins and is often found in coastal areas and scrub covered gullies. It climbs into the medium and high canopy where it forms heavy, long lived masses which smother trees and shrubs.

Section 72(1)(c) The medium score from the weed risk assessment reflects the fact that Madeira vine is a weedy species in an early stage of invasion in Canterbury but because it does not produce seed its spread is less likely to be rapid and widespread. Early intervention could probably eradicate this weed in Canterbury.

The potential serious adverse effects provide justification for the inclusion of Madeira vine in the RPMS (Section 72(1)(c)).

55 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Madeira vine (Anredera cordifolia)

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 3 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 4 numerous small (2), numerous large (0)

C. Reproduction 1 Species cryptic and cannot be detected before it reproduces 1 or 0 1 2 Produces viable seed 2 or 0 0 3 Seed dispersed primarily by: small birds, wind, or water (2), 2 or 1 0 large birds or passive/accidental dispersal (1) 4 Minimum generation time < 3 years (2), > 3 years (1) 2 or 1 2 5 Persistent vegetative organs above or below ground, 2 or 0 2 or seed bank (> 1 year) 6 Juveniles common within 100 m parents 1 or 0 0 Spread score (sum B1–3 + C1–6) 12 Impacts x spread score 96

56 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

15. False tamarisk (Myricaria germanica)

False tamarisk is a deciduous shrub which grows up to 1.5m high. It is only known in New Zealand from Canterbury braided rivers (Heenan et al.1999). It is thought to have been introduced to combat soil erosion or mistakenly thought to be Tamarix (which looks very similar). There are no reports of it being cultivated or adventive in New Zealand (Sykes and Williams 1999). However, it is now listed on the NPPA.

False tamarisk has flourished on disturbed sites where the river channels are constantly changing in flood events. In its native Europe it occupies similar disturbed sites in riverbeds. Ironically it has become rare in Europe where human interference with the natural flow of river systems has stabilised the river channels. At these sites false tamarisk is replaced by willows.

False tamarisk has, so far, only been recorded from the braided riverbeds of Canterbury. Dispersal seems to be mainly from vehicles carrying seed, as new sites are often near vehicle access points to riverbeds eg. boat launch areas.

Initially the Department of Conservation (DoC) tried to eliminate false tamarisk from riverbeds. Glyphosate worked best and large infestations in riverbeds were sprayed by helicopter. However it left bare swathes which were reinvaded with other weeds and the collateral damage was too great. DoC now controls false tamarisk only near areas of high conservation value (Helen Braithwaite, pers.comm.).

Current and potential habitat invasion False tamarisk is widespread in Canterbury braided rivers including the Ashley (Lees Valley), Waimakariri, Rakaia and Rangitata. It appears to be spreading rapidly inland probably due to seed being carried on boats and the vehicles towing them.

There is a high risk that false tamarisk will invade the headwaters of the braided river systems and inland basins of Canterbury eg. the Heron and McKenzie Basins.

Risks [under Section 72(1)(c)] False tamarisk is considered to have economic and conservation impacts. • Economic impacts Although DoC now only controls areas of high natural values there is still a cost associated with this. • Conservation impacts False tamarisk can form dense stands which could inhibit other plants especially small indigenous riverbed species. Like other weeds that invade braided riverbeds, false tamarisk is likely to reduce the habitat available for birds that require open stony riverbeds for nesting. These woody weeds also provide cover for predators of these birds.

Section 72(1)(c) The high score in the weed risk assessment is because false tamarisk has only recently naturalised (1999) and therefore has a high score for the invasion stage. Usually weeds at an early stage of invasion are the easiest to control but in this case colonisation of the braided river beds has been so rapid that this opportunity has been lost. Control is now for high value conservation areas only.

Although false tamarisk is listed on the NPPA it is ranked as low priority as it is unlikely that it would be cultivated or sold in nurseries.

57 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

The serious adverse effects of false tamarisk are such that inclusion in the RPMS is justified.

Weed Risk Assessment

False tamarisk (Myricaria germanica)

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 9 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 2 numerous small (2), numerous large (0)

58 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

16. Royal fern (Osmunda regalis)

Description Royal fern is a deciduous fern with rhizomes that form a short woody trunk up to 1.5m high. Plants die back to the woody trunk in winter. It is locally abundant in Waikato/Bay of Plenty but sparse or absent elsewhere and is not known to be present in the South Island apart from two unconfirmed sites in Christchurch. It is thought to be a potential major weed of peatlands throughout New Zealand, especially after disturbance events such as fire.

Spread is mainly by wind dispersed spores in areas where the fern is well naturalised. The spores are viable for one week which is a limiting factor. There is also likely to be some spread by human activities.

Royal fern is on the NPPA list which should reduce long-distance spread to regions not yet invaded.

Current and potential habitat invasion Although Royal fern is a serious weed in parts of the North Island it is currently not known from Canterbury apart from two unconfirmed sites in Christchurch. Two observations of Royal fern were made during the Christchurch Rivers Ecological Assets Survey (CREAS) where all streams and drains in the Christchurch area were surveyed for their ecological values (Manfred von Tippelskirch, pers.comm.). However, there is no record of Royal fern on the database storing all the data from the CREAS survey (Trevor Partridge, pers.comm.).

Should royal fern become established in Canterbury there is potential for it to invade swamps, wet ground, stream sides and drains throughout Canterbury.

Risks [under Section 72(1)(c)] Royal fern is considered to have potential economic and conservation impacts. • Economic impacts There are currently few recordings of Royal fern in Canterbury. Should it become established and naturalised in Canterbury then there would be costs involved in trying to eliminate it. • Conservation impacts Its dense growth habit is a threat to indigenous species including endangered biota and habitat quality. Royal fern can naturalise and form dense colonies in a range of wetland types in New Zealand, especially in disturbed areas and under the shade of willows. They can displace other small wetland plants.

Section 72(1)(c) The present low (or zero) number of sites is reflected in the high score in the weed risk assessment. Royal fern is a serious weed in the Waikato / Bay of Plenty area and it is important that any plants are eliminated before it becomes a conservation and amenity threat to the wetlands and waterways of Canterbury.

Royal fern is assessed as having potentially serious adverse effects should it become established in Canterbury, so inclusion in the RPMS is justified.

59 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Royal fern (Osmunda regalis)

60 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

17. Asiatic knotweed (Reynoutria japonica)

Description Asiatic knotweed is an upright shrub-like herbaceous perennial that can rapidly grow to 2m in height. It has reddish shoots that become hollow at maturity. It is very similar to giant knotweed but giant knotweed is taller and does not produce mature fruit in New Zealand. Both species have hybrids which are often intermediate between the 2 parents. Both Asiatic and giant knotweed are garden escapes. Variety 'Compacta' has been widely grown in New Zealand. Both species are listed on the NPPA which should stop its spread to other catchments.

Asiatic knotweed grows rapidly and extensively from rhizomes and multiple stems. It produces relatively long-lived and well dispersed seed and tolerates wet to moderately dry conditions, warm to cold temperatures, but is intolerant of shade.

Distribution of Asiatic knotweed is limited by lack of dispersal ability but could become much more widespread especially in high rainfall areas. It invades shrubland and riparian areas and is difficult to control due to the large underground biomass.

Current and potential habitat invasion Asiatic knotweed has a scattered distribution in the North Island is in the upper half of the South Island. It is locally abundant in Westland. In Canterbury it is only known from near Christchurch where it has probably escaped from gardens.

Dumping of garden rubbish is the most likely method of spreading this weed in Canterbury, so areas around population centres are most likely at risk. Should Asian knotweed become more widespread in Canterbury it is likely to naturalise on roadsides, waste places and riverbanks.

Risks [under Section 72(1)(c)] Asian knotweed is considered to have economic, conservation and health impacts. • Economic impacts Should Asian knotweed become more widespread then it is likely to have costly economic impacts due to the cost of controlling the weed. As Asian knotweed can be an aggressive coloniser of rough pasture there is likely to be an economic impact on farmers due to reduced grazing. As Asian knotweed is at an early stage of invasion in Canterbury prompt intervention in controlling or eliminating this weed would reduce the potential economic impacts (Harris and Timmins, 2009). • Conservation impacts Asiatic knotweed forms dense stands that shade and crowd out all other vegetation. It is an aggressive coloniser of disturbed areas and riparian zones. Its dense growth habit is a threat any indigenous biota (plants, birds, fish and insects) inhabiting the riparian zone. The dense thickets also prevent native plants establishing and competes with other native flora eg.flax and raupo.

• Health and recreation impacts This plant may cause contact dermatitis. As Asian knotweed may establish on river banks and the access to them it may cause some irritation and discomfit to recreational users.

Section 72(1)(c)

61 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Asian knotweed has potential to be a serious weed in Canterbury. Early intervention at this stage should prevent Asian knotweed from becoming established in new areas.

The potential serious adverse effects provide justification for the inclusion of Asian knotweed in the RPMS (Section 72(1)(c).

Weed Risk Assessment

Asiatic knotweed (Reynoutria japonica)

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 2 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 4 numerous small (2), numerous large (0)

62 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

18. Giant knotweed (Reynoutria sachalinensis)

Description Giant knotweed is an upright shrub-like herbaceous perennial that can rapidly grow to 4m in height. It is very similar to Asiatic knotweed but giant knotweed is taller and does not produce mature fruit in New Zealand. Both species have hybrids and which are often intermediate between the 2 parents.

Giant knotweed is a sporadic garden escape usually found on roadsides near population centres. The plant can form conspicuous giant clumps on roadsides. Like Asian knotweed it is difficult to control due to the large underground biomass. Giant knotweed grows extensively from rhizomes and multiple stems, it can tolerate wet to moderately dry conditions and warm to cold temperatures, but is intolerant of shade. It spreads mainly by pieces of rhizome sprouting after being dumped or moved.

Current and potential habitat invasion Giant knotweed has a scattered distribution throughout New Zealand but is only common in Westland. Distribution is limited by lack of dispersal ability (due to lack of mature fruit) but it could become much more widespread especially in high rainfall areas. In Canterbury it is only known from Christchurch and Hororata.

Dumping of garden rubbish is the most likely method of spreading this weed in Canterbury, so areas around population centres are most likely at risk. Should giant knotweed become more widespread in Canterbury it is likely to colonise road edges and waste places.

Risks [under Section 72(1)(c)] Giant knotweed is considered to have potential economic and conservation impacts. • Economic impacts Should giant knotweed become more widespread then it is likely to have costly economic impacts due to the cost of controlling the weed. As giant knotweed is at an early stage of invasion in Canterbury prompt intervention in controlling or eliminating this weed would reduce the potential economic impacts (Harris and Timmins, 2009).

• Conservation impacts Giant knotweed forms dense stands that shade and crowd out all other vegetation. It is an aggressive coloniser of disturbed areas and waste places. It forms dense, long-lived thickets, shading out other species and preventing native seedlings from establishing.

Section 72(1)(c) Giant knotweed has potential to be a serious weed in Canterbury. Early intervention at this stage should prevent giant knotweed from becoming established in new areas.

The potential adverse effects provide justification for the inclusion of giant knotweed in the RPMS (Section 72(1)(c).

63 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Giant knotweed (Reynoutria sachalinensis)

C. Reproduction 1 Species cryptic and cannot be detected before it reproduces 1 or 0 1 2 Produces viable seed 2 or 0 0 3 Seed dispersed primarily by: small birds, wind, or water (2), 2 or 1 2 large birds or passive/accidental dispersal (1) 4 Minimum generation time < 3 years (2), > 3 years (1) 2 or 1 2 5 Persistent vegetative organs above or below ground, 2 or 0 2 or seed bank (> 1 year) 6 Juveniles common within 100 m parents 1 or 0 1 Spread score (sum B1–3 + C1–6) 14 Impacts x spread score 91

64 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

19. African club moss (Selaginella kraussiana)

Description African club moss is a small carpet-forming fern ally groundcover. It forms loose mats by the creeping stems rooting at the nodes and radiating out to form large patches. It looks similar to many native mosses, liverworts and Leptinella species. It grows on the ground or epiphytically and can disperse widely and quickly. It can tolerate hot and cold temperatures and light to deep shade but requires reasonably damp soils.

Dispersal is by spores and stem fragments via boots, livestock, water movement, dumped vegetation and potted plants. African club moss has been sold commercially in the past so it can be found scattered throughout New Zealand. It is now included in the NPPA so it cannot be sold, propagated or distributed.

Current and potential habitat invasion African club moss invades disturbed forest, shrubland, stream sides and fernland. It is widespread on lowland sites throughout New Zealand. In Canterbury it is known from Christchurch and South Canterbury and is probably quite common in gardens and pot plants.

African club moss has the potential to invade damp forest floors and stream banks in lowland forest throughout Canterbury.

Risks [under Section 72(1)(c)] African club moss is considered to have economic and conservation impacts. • Economic Should African club moss become more widespread throughout Canterbury it is likely to have economic impacts due to the cost of control. • Conservation African club moss creeps over the forest floor inhibiting the recruitment of native plants. The lack of native understorey allows more aggressive weeds, especially vines, to establish. Although African club moss is a small plant it can lead to serious invasion by vines.

Section 72(1)(c) The high score from the weed risk assessment reflects the fact that African club moss is a serious weed but is at an early stage of invasion. African club moss should be considered for eradication in Canterbury as the small populations at a limited number of sites means that eradication is realistic and feasible.

The potential adverse effects provide justification for the inclusion of African club moss in the RPMS (Section 72(1)(c).

65 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

African club moss (Selaginella kraussiana)

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 1 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 4 numerous small (2), numerous large (0)

66 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

20. Grey willow (Salix cinerea)

Description Grey willow is a deciduous tree which can grow up to 7m in height. Grey willow trees are either male or female with the female trees producing many short-lived seeds that are widely dispersed. Hence control measures may target female trees first as they are the seed producers.

Grey willow is a very serious weed of wetlands and waterways throughout New Zealand. It forms a dense canopy and completely alters the ecology in wetlands. It spreads mainly from seeds but also suckers forming dense thickets. It grows rapidly and appears to thrive in a wide range of environments. It can tolerate flooding, semi-shade and a wide range of temperatures.

In the past grey willow has been sold in nurseries throughout New Zealand, and is known to be cultivated in gardens in cooler districts of the South Island. In many wetland areas it has been purposely planted. It is now listed on the NPPA which should prevent spread to new regions.

Current and potential habitat invasion Grey willow is locally abundant in some areas but significant areas of New Zealand are either free of this plant, or with limited populations. In Canterbury it is widespread and abundant around lakes, rivers and swamps in the high country, foothills, plains and near population centres. The wide environmental tolerances, high seed production and ability to form dense thickets means that grey willow has the widest geographical distribution of any weed in Canterbury. It is present in waterways from sea level into the foothills and far into the Southern Alps close to the main divide.

Left unchecked grey willow would continue to invade wetlands, rivers and swamps throughout Canterbury.

Risks [under Section 72(1)(c)] Grey willow is considered to have economic and conservation impacts. • Economic impacts Because grey willow is so widespread and invasive the cost of control is high. • Conservation impacts Grey willow is the greatest threat to wetlands in New Zealand, due to its tall stature and tolerance of a range of soils and flooding. It causes major changes to wetland processes at invaded sites. Original native wetlands in Canterbury had no trees – wetlands were open communities supporting low stature light demanding plants (rushes, sedges and small herbs). When willows were planted they changed the whole ecology of the wetlands by excluding light and thereby excluding all the native light demanding species. Willows also change the whole drainage system by causing blockages, flooding and structural changes in the waterway.

67 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Section 72(1)(c) The long time that grey willow has been established in New Zealand and the numerous large infestations in Canterbury is reflected in the relatively low score in the weed risk assessment. To prevent ongoing spread of grey willow considerable resources will be needed. However the very serious impacts of grey willow on wetlands means that control is a high priority especially in areas where there are high conservation values and willows are not yet dominant.

The very serious adverse effects, as described here, provide justification for the inclusion of grey willow in the RPMS.

Weed Risk Assessment

Grey willow (Salix cinerea)

C. Reproduction 1 Species cryptic and cannot be detected before it reproduces 1 or 0 0 2 Produces viable seed 2 or 0 2 3 Seed dispersed primarily by: small birds, wind, or water (2), 2 or 1 2 large birds or passive/accidental dispersal (1) 4 Minimum generation time < 3 years (2), > 3 years (1) 2 or 1 1 5 Persistent vegetative organs above or below ground, 2 or 0 2 or seed bank (> 1 year) 6 Juveniles common within 100 m parents 1 or 0 1 Spread score (sum B1–3 + C1–6) 9 Impacts x spread score 72

68 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

21. Japanese spindle tree (Euonymus japonicus)

Description Japanese spindle tree is an evergreen shrub or small tree which can grow up to 7m in height. It has brightly coloured ornamental fruits which makes it attractive to birds. It has been sold in the past as a desirable garden plant with brightly coloured fruit and variegated foliage. Japanese spindle tree is widely cultivated often as a variegated form which, in time, appears to revert to the green form. All naturalised material appears to be non-variegated.

Japanese spindle tree is mainly spread by birds eating the seed and distributing it widely. Physical control is by hand, pulling seedlings or cutting and stump treating larger trees. Chemical control of larger shrubs and trees is by cutting and applying herbicide to cut stumps.

Japanese spindle tree is found in scrub, regenerating forest, sand dunes and waste places and thrives in both open and shady places. Overseas it is naturalised and invasive in a number of north-western American states.

Current and potential habitat invasion Currently the naturalised distribution of Japanese spindle tree is confined to coastal and lowland areas of the North Island. It is well-established in Auckland and also has been collected from Levin and Pukerua Bay. It is not regarded as naturalised in Canterbury (Mahon, 2007) but is known from Christchurch gardens and is starting to be seen in the wild around population centres eg. behind Lyttelton, Hamner and Timaru (Alan McDonald, pers. comm.). Large infestations under plantation forests at Chaneys/Bottle Lake north of Christchurch have recently been noticed (Trevor Partridge, pers.comm.). There is also a recording between Lakes Pukaki and Ohau (Ecan).

Japanese spindle tree has potential to spread further than its current distribution. It is likely to move out from town gardens to become fully naturalised in Canterbury unless it is controlled. It is likely to invade scrub and regenerating forest throughout Canterbury. Coastal shrublands and sand dunes are also at risk.

Risks under [Section 72(1)(c)] Japanese spindle tree is considered to have economic, conservation and health impacts. • Economic impacts Should Japanese spindle tree become more widespread then there are likely to be economic impacts due to the cost of control. • Conservation impacts Japanese spindle tree invades forest margins, disturbed and secondary forest and shrubland, coastal scrub and forest, cliffs and sand dunes. It forms dense stands in open or shady margins preventing the establishment of native plants. • Health impacts Japanese spindle tree is toxic to humans and animals (Connor, 1977). All Euonymus spp. are believed to be highly toxic due to the presence of glycosides. Ingestions of these plants, especially in large amounts, are likely to cause serious effects to major body organs such as the liver, heart or kidneys. All parts of these plants are poisonous and ingestion will lead to vomiting, diarrhea, weakness, chills, coma and convulsions.

69 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Section 72(1)(c) Japanese spindle tree has potential to become a serious weed in Canterbury and this is reflected in the high score in the weed risk assessment. Early intervention at this stage should prevent it from becoming fully naturalised in Canterbury.

The potential serious adverse effects provide justification for the inclusion of Japanese spindle tree in the RPMS (Section 72(1)(c).

Weed Risk Assessment

Japanese spindle tree (Euonymus japonicus)

B. Invasion stage 1 10 - (naturalisation decade) 10 to 0 7 2 Recently (< 5 yrs) recognised as weed 2 or 0 0 3 No./size infestations: one small (8), several small/single large (4), 8 to 0 4 numerous small (2), numerous large (0)

70 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

22. Pigs ear (Cotyledon orbiculata)

Description Pigs ear is a clump forming succulent species with thick grey leaves. It has clusters of orange bell shaped drooping flowers. Its country of origin is South Africa. It is able to thrive in hot sunny conditions on dry rock with very little soil. It is invasive on coastal slopes and beaches, steep banks, rocky outcrops, cliff faces and bare ledges. It sometimes grows in low scrub and dry depleted grassland (Birdlings Flat). Pigs ear is one of a suite of weeds that are a threat to some rare and threatened plants of rock outcrops. Other weedy species that appear to thrive in dry rocky environments include boneseed, spur valerian, fennel, polypodium fern, wallflower, hawthorn and boxthorn.

Being a hardy succulent, pigs ear is popular as a pot plant and in gardens particularly in hot dry places. It is now listed on the NPPA but in the past it has often used for landscaping and is still swapped or traded on Trade Me and street markets. Seed is spread by wind and the thick leaves will root in suitable conditions.

There is little known about an effective herbicide for pigs ear. Control is by hand pulling plants which may be difficult in inaccessible areas on rock bluffs.

Current and potential habitat invasion Pigs ear occurs in dry rocky places throughout New Zealand. In the South Island it is present on the Marlborough Coast, Canterbury and the Otago Peninsula. In Canterbury it grows along the North Canterbury coast, the Port Hills and Banks Peninsula on cliffs, rock outcrops, banks and dry grassland.

It is likely the pigs ear will continue to invade dry rocky places and dry grassland in lowland and coastal areas of Canterbury.

Risks under [Section 72(1)(c)] Pigs ear is considered to have economic, conservation and health impacts. • Economic Because pigs ear is difficult to control the cost of control is high. The current manual method of removing pigs ear is time consuming and expensive especially when it grows on inaccessible cliff faces. Control may have to be confined to areas of high conservation value. • Conservation Pigs ear competes with and replaces native vegetation including threatened plants especially on rock outcrops where niche sites are limited, eg. the rare blanket fern Pleurosorus rutifolius on the dry sunny rock outcrops of the Port Hills. Other plants that are threatened by pigs ear include the iconic prostrate kowhai (Sophora prostrata) and several regionally endemic species such as the Banks Peninsula hebe (Heliohebe lavaudiana), Banks Peninsula blue tussock (Festuca actae) and Banks Peninsula hebe (Hebe strictissima). Pigs ear competes with a suite of other indigenous rock outcrop plants. Of note are the ‘hot rock ferns’ so named because of their unusual ability to survive hot dry conditions when most ferns prefer moist shady conditions. Hot rock ferns include Pleurosorus rutifolius, Pellaea calidirupium, Cheilanthes sieberi, Cheilanthes distans, Asplenium appendiculatum and Asplenium flabellifolium.

• Health Pigs ear leaves can be used to treat corns and fever blisters and the warmed leaf juice can be used as drops for earache. It may also be applied in the form of a hot poultice to treat boils, earache, inflammation or warts. Internal use is dangerous and

71 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

potentially lethal, and the toxicity is affected by the moisture content of the leaves. There have been no recorded incidents of this plant causing harm to humans but there have been cases in California where sheep have died when fed pigs ear and all Cotyledon species should be considered poisonous. In South Africa, the disease caused by eating these plants, called cotyledonosis, has poisoned sheep and goats but rarely other animals. Ranchers in South Africa found that the meat of animals killed by cotyledonosis remains toxic.

Section 72(1)(c) Pigs ear is a serious weed and is a threat to the special indigenous plants that occupy the same dry rocky environments. To control and prevent ongoing spread of pigs ear considerable resources will be needed. However the very serious impacts of pigs ear on threatened plant communities means that control is a high priority especially in areas where there are high conservation values.

The serious adverse effects provide justification for the inclusion of pigs ear in the RPMS [Section 72(1)(c)].

72 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Pigs ear (Cotyledon orbiculata)

D. Cultivation and perceptions 1 Present as: mass plantings (3), frequent smaller plantings (2), 3 to 0 1 infrequent small plantings (1), not planted (0) 2 No. nurseries selling species: > 3 ,< 3, 3 to 0 1 3 Is it a crop plant? 1 or 0 0 4 Does it have unpleasant features? 1 or 0 1 Public attitudes score (sum D1–3 minus D4) 1

73 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

23. Cotoneaster simonsii

Description Cotoneaster simonsii is a deciduous or evergreen shrub that can grow up to 4m high. It has small flowers but bright shining orange berries that are very attractive to birds. It can form dense stands which out-compete native species in a wide range of habitats. It is very tolerant of drought, a wide range of temperatures and is shade tolerant. It is long-lived, matures quickly and produces lots of viable seeds which are widely distributed by birds.

Cotoneaster simonsii is a garden escape and has been used for hedges and shelter in the past. It is now listed as an NPPA species.

Current and potential habitat invasion Cotoneaster simonsii is well established in Canterbury from Banks Peninsula to the foothills and the main divide and from North to South Canterbury. On the Amuri Range south of Hanmer it has demonstrated that it is a major part of the flora and is shade tolerant enough to replace Coprosma rhamnoides shrubs in the understorey (Miles Giller, pers comm.). It is found in the Waitaki Valley, at Mt Cook, Ashburton, mid Canterbury, Banks Peninsula and forms the dominant cover under some conifer plantations at Hanmer.

The ability of Cotoneaster simonsii to invade dry cool forest up to 900m, bluffs, rock outcrops, slips and riverbeds means that it could potentially invade these environments throughout Canterbury.

Risks [under Section 72(1)(c)] Cotoneaster is considered to have economic and conservation impacts. • Economic impacts The cost of controlling Cotoneaster simonsii is costly. It is currently too widespread in Canterbury to eradicate so it is probably more cost effective to try to prevent new areas of infestation rather to try controlling it everywhere. • Conservation impacts Being shade tolerant Cotoneaster simonsii can form a smothering understorey in open forest which prevents recruitment of native species. It invades forest margins and can overtop and replace shrubs. The wide environmental tolerances of Cotoneaster simonsii, its prolific seed production and ability to form dense stands means that it is a serious weed of native forest, shrublands and open environments.

Section 72(1)(c) The high score from the weed risk assessment recognises that Cotoneaster simonsii is a serious weed well established in Canterbury. To prevent ongoing spread of Cotoneaster simonsii considerable resources will be needed. However the very serious impacts of Cotoneaster simonsii in regenerating forest and shrublands means that control is a high priority especially in areas where there are high conservation values.

The very serious adverse effects, as described here, provide justification for the inclusion of Cotoneaster simonsii in the RPMS (Section 72(1)(c).

74 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Cotoneaster simonsii

75 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

24. Puna grass (Achnatherum caudatum)

Description Puna grass is a tall tussock-like grass that grows up to 1m tall. The flowerheads look similar to nassella tussock. Like nassella, South America is the country of origin. Puna grass is a weed of grasslands and riparian vegetation, as well as waste areas and roadsides. In Australia it is considered to be similar in threat-potential to nassella tussock. Puna grass is not particularly palatable to stock. It is spread by seed and is difficult to control once established.

Current and potential habitat invasion Puna grass is known only from Canterbury. One site is on a farm near Amberley and the other is in a paddock in Christchurch. Both sites are in sandy soils. Ecan is controlling the 2 known populations by grubbing the plants.

There is potential to spread in sandy soils in coastal areas.

Risks [under Section 72(1)(c)] Puna grass is considered to have economic, agricultural and conservation impacts. • Economic impacts Should puna grass escape from the current 2 locations there are the potential costs of control that could be on a similar scale to the control of nassella tussock. • Agricultural impacts Puna grass occupies similar habitats to nassella so the impact would be similar should puna grass escape from its current locations. As puna grass is not palatable it is likely to displace other more palatable species. Seeds are likely to be transported to different areas by stock. • Conservation impacts Should puna grass become more widely established it is likely to change the vegetation composition of native ecosystems in short tussock grassland and sandy soils along the coast.

Section 72(1)(c) Puna grass has the potential to become a serious weed if it becomes more widely established in Canterbury. The current very localised sites of invasion and short time since naturalisation is reflected in the high score in the weed risk assessment. Puna grass is a high priority for total control with the aim of eradication of this species from Canterbury and New Zealand.

The potential serious adverse effects provide justification for the inclusion of puna grass in the RPMS (Section 72(1)(c).

76 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Puna grass (Achnatherum caudatum)

77 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

25. Russell lupin (Lupinus polyphyllus)

Description Russell lupin is a perennial that can grow up to 1.5 metres tall. It flowers and sets seed in the summer and dies back to the stem base over winter. Russell lupins produce long, colourful flower heads. The flowers are pea-like and come in a variety of colours. Stout seedpods are produced that explode in the summer heat, releasing many dark brown seeds.

Although Russell lupins have attractive flowers, they can be an aggressive weed. Of particular concern is the invasion of Russell lupins into Canterbury’s braided riverbeds, and the impacts they have on these ecosystems. Lupins are well adapted to living in the challenging environments of braided rivers. They can produce their own nutrients (nitrogen) and are very effective at dispersing their seeds. The seeds are dropped close to the parent plant, allowing the population to spread a couple of metres each year. Seeds also spread further if they are carried in waterways, allowing Russell lupins to creep down riverbeds and invade new areas. Lupin spread is also increased by people actively dispersing seeds along roadsides and waterways, and by roadwork contractors using gravel containing seed. Russell lupins may appear harmless and pretty growing in the garden or along roadsides, but the potential for them to escape and take over nearby waterways is enormous.

Russell lupin seeds are still sold in many nurseries and tourist shops. The attractive displays along the roadsides in the McKenzie Basin are a highlight for tourists and travelling New Zealanders alike. This is problematic for getting the message across that they are detrimental to conservation values.

Russell lupins are not listed on the NPPA probably because of their popularity as a garden plant and attraction for tourists.

Current and potential habitat invasion Russell lupins can be aggressive weeds and are invading Canterburys braided riverbeds. They have spread along roadsides in the McKenzie Basin, Athurs Pass and Mt Cook and because of the attractive flowers they have become popular with tourists. They are spread along braided rivers, streams and riverbeds from lowland sites to montane and subalpine areas eg Rangitata, Mt Cook, Ahuriri.

There is potential for the Russell lupin to spread further along roadsides and riverbeds.

Risks [under Section 72(1)(c)] Russell lupin is considered to have mainly conservation impacts. • Conservation Canterbury’s braided rivers are home to unique native plant communities. Special plants such as the cushion-forming forget-me-not (Myosotis uniflora) and rare, tiny woodrush (Luzula celata) are mostly confined to riverbeds. Whole plant communities are especially adapted to growing in the challenging environment of shifting gravels, extreme temperatures and limited nutrients. This natural vegetation is often low-lying and sparse, leaving plenty of room for Russell lupins to move into. Dense stands of lupins eventually shade out and displace these special threatened plants and whole native plant communities.

Unique birds live and breed in the braided riverbeds of Canterbury. Birds such as the vulnerable wrybill and black-fronted tern have adapted to nesting and feeding in unstable braided river environments. One of the world’s rarest wading birds, the black stilt, also feeds in shallow river braids. Russell lupins change these unstable

78 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

braided river environments by forming dense stands on the bare gravel areas. Their roots become entwined and hold the gravel together, forming stable areas. The river erodes the edges, forming steep banks which drop into deep, fast-flowing channels, unsuitable for wading birds to feed in. The dense stands also take over the open spaces braided river birds like to nest in.

Section 72(1)(c) Russell lupin is a weed that has very serious ecological repercussions on the braided river ecosystems and montane natural areas of Canterbury. It has a serious effect on the viability of threatened species, the survival and distribution of indigenous plants and animals, and the sustainability of ecological processes and biological diversity. Its serious adverse effects provide justification for the inclusion of Russell lupin in the RPMS (Section 72(1)(c).

79 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Russell lupin (Lupinus polyphyllus)

D. Cultivation and perceptions 1 Present as: mass plantings (3), frequent smaller plantings (2), 3 to 0 2 infrequent small plantings (1), not planted (0) 2 No. nurseries selling species: > 3 ,< 3, 3 to 0 3 3 Is it a crop plant? 1 or 0 0 4 Does it have unpleasant features? 1 or 0 0 Public attitudes score (sum D1–3 minus D4) 5

80 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

26. Boxthorn (Lycium ferocissimum)

Description Boxthorn is a densely branched spiny evergreen shrub from South Africa. It can grow up to 6m tall and has orange red berries.

Boxthorn is an aggressive coloniser of sand dunes, gravel, coastal pasture, scrub and waste places. The spiny nature of boxthorn makes it suitable for stock proof barriers and windbreaks and it has been used for this purpose in the past. Boxthorn rarely spreads into good pasture but can be a troublesome weed in some rough country. Because of its tolerance to salt spray and its ability to grow on unstable sand dunes it is often the only woody plant present on some coastal sites. Boxthorn berries are spread by birds from farm hedges and waste places.

Boxthorn tolerates a wide variety of soil types (sand to rocky cliffs), drought, salt, wind and a wide range of temperatures. It is a long-lived shrub that forms dense tall stands, excluding most other vegetation. Boxthorn is not very palatable so grazing by stock is unlikely to control it. Fire is not an effective method of control. Glyphosate is the most effective means of controlling boxthorn. Cut bushes can coppice so need to be treated.

Current and potential habitat invasion Boxthorn is scattered throughout New Zealand. In Canterbury it of concern on coastal cliffs (Motanau), rock outcrops on farmland (Waipara Gorge, Weka Pass, Banks Peninsula), inland dry rocky places (Rakaia Gorge, Waitaki Valley) and dunelands (Kaitorete Spit).

Boxthorn is likely to continue to spread into dry rocky areas from the coast to inland gorges. Of particular concern is spread onto limestone outcrops where there are valuable threatened plant communities.

Risks [under Section 72(1)(c)] Boxthorn is considered to have economic and conservation impacts. • Economic impacts Control of boxthorn on farmland and conservation land is likely to have costly economic impacts. As boxthorn is widespread in Canterbury control should be focussed on high value areas eg rock outcrops. • Conservation impacts Boxthorn is an aggressive coloniser of open areas, particularly rock outcrops, sand dunes and coastal cliffs. It can overtop native plants, excluding light and absorbing water and nutrients to the detriment of the native species. Limestone rock outcrops may harbour threatened plants. Heliohebe maccaskillii and Gentianella calcis ssp. waipara are nationally threatened species and occur only on some North Canterbury limestone outcrops.

Box thorn may entrap petrels and other seabirds that become entangled in the thorny trees.

Section 72(1)(c) Boxthorn has been naturalised for a long time and this is reflected in the medium score in the weed risk assessment. However, boxthorn is a serious weed in areas of high conservation value where threatened plants may be present (eg. limestone rock outcrops). Site led control in areas of high conservation value is required.

81 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

The potential serious adverse effects provide justification for the inclusion of boxthorn in the RPMS [Section 72(1)(c)]. Weed Risk Assessment

Boxthorn (Lycium ferocissimum)

82 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

27. Common polypody (Polypodium vulgare)

Description Common polypody is a small hardy evergreen fern. It has a creeping rhizome but probably spreads mainly by its spores. Common polypody was first recorded as naturalised in New Zealand on the Port Hills near Lyttelton during the 1960s and 1970s, but was probably present earlier (Lovis, 1980). Lovis’s (1980) description of the distribution of Polypodium vulgare on the Port Hills suggests it was rather localised at that time, but it is now widespread, occurring from Godley Head to Gebbies Pass. It is also known from coastal hillsides in the Wellington area (Hongoeka Bay).

Polypodium vulgare has been cultivated in New Zealand, but apparently not to any great extent. In the past, it has been promoted for planting in New Zealand (Van der Mast and Hobbs (1998). Currently it does not appear to be sold in nurseries but it is likely that plants from home gardens are exchanged or sold at fairs and garage sales.

There are no native species of the genus Polypodium in New Zealand but common polypody belongs to the same family as one of our common native ferns, hound’s tongue (Microsorum pustulatum subsp. pustulatum).

Common polypody is an aggressive spreading fern. It spreads easily via a network of spreading, mat-like rhizomes and spores. It has a wide habitat ranging from open sunny outcrops to the shaded understorey of a forest. Canterbury’s rock outcrops are sensitive areas and, in these situations, common polypody can compete with threatened native vegetation.

To date it is difficult to control and control techniques are being explored (Di Carter, pers.comm.).

Current and potential habitat invasion Common polypody is currently known only from coastal cliffs near Wellington and from Canterbury. In Canterbury it is abundant and widespread along the Port Hills of Christchurch and it has recently been recorded on coastal to montane greywacke rock outcrops in North Canterbury (Miles Giller, pers. comm.).

Because common polypody is spread by spores in the air it has spread rapidly from the Port Hills into North Canterbury and may be more widespread. The potential areas under threat include dry and damp sites on rock outcrops, bluffs, forest understorey and shrublands throughout Canterbury.

Risks [under Section 72(1)(c)] Common polypody is considered to have serious conservation impacts. • Conservation impacts Common polypody is a very significant threat to small and sparse populations of rare plants on rock outcrops. It occupies a full range of habitats from dry sunny sites to shady wet mossy sites on rock outcrops and cliffs. These are the specific habitats of several threatened plants. On the Port Hills rock outcrops it is competing with rare and threatened plants including the rare blanket fern (Pleurosorus rutifolius) and Lyttelton forget-me-not (Myosotis australis var. lytteltonensis), the iconic prostrate kowhai (Sophora prostrata) and several regionally endemic species such as the Banks Peninsula hebe (Heliohebe lavaudiana), Banks Peninsula blue tussock (Festuca actae), Banks Peninsula hebe (Hebe strictissima) and a diverse range of other rock outcrop plants. As it is also shade tolerant common polypody can also affect the forest understorey structure and prevent regeneration.

83 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Section 72(1)(c) Common polypodium poses a significant threat to several rare indigenous plants on rock outcrops as it occupies a similar ecological niche. It is spreading rapidly with no current means of control apart from manually pulling it out. This may be the only option and should be undertaken in high value areas such as where threatened plants are present. Limiting distribution of common polypody would prevent further spread and help to protect forest and rock outcrop values throughout Canterbury.

The high score from the weed risk assessment reflects that common polypody is a serious weed with high impacts on threatened plants and indigenous biodiversity. Site led control in areas of high conservation value and prevention of sale, distribution and propagation is required.

The potential very serious adverse effects provide justification for the inclusion of common polypody in the RPMS [Section 72(1)(c)].

84 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Common polypody (Polypodium vulgare)

85 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

28. Carex pendula

Description Carex pendula is a tall, exotic, shade tolerant, perennial sedge which grows in damp areas. It is the tallest sedge growing in New Zealand with stems up to 2.5m long. It has distinctive drooping flower spikes crowded with many seeds. This large sedge looks similar to some native sedges Carex lessoniana and Carex geminata but the native sedges are smaller (to 1.5m tall). Carex pendula is typically found on river banks but appears to thrive in other wetland areas and in open forest.

The first New Zealand collection of Carex pendula was in 1962 at Otahuna near Tai Tapu (Healy & Edgar, 1980). It is a garden escape and is classified as fully naturalised (Mahon, 2007). Although it is native to Europe, Asia and North Africa there are many reports of it recently expanding its range in England. It is sold as an ornamental plant but gardeners are discovering its weedy tendencies. It matures and sets seed rapidly with thousands of seeds germinating. It is not indigenous to Ireland but thrives there and is known to aggressively spread into pasture and forest (Alan MacDonald, pers. comm.).

In response to concern expressed by volunteer groups working in Ernle Clark Reserve on the Heathcote River, Ecan agreed to survey the Avon and tributaries and consider targeting Carex pendula for eradication. The Christchurch City Council is cutting Carex pendula so that it doesn’t seed, as part of the maintenance of river banks. A literature search on Carex pendula has been completed (Hazel Gatehouse, 2009).

Current and potential habitat invasion Carex pendula is currently known from Waimairi Stream (a tributary of the Avon)(Ecan survey), the Heathcote River and the Halswell River near Tai tapu. There are also sightings from Rangiora and Ashburton. In the last few years it has spread down the Heathcote River from opposite Princess Margaret Hospital. Volunteers working in Ernle Clark Reserve have noticed Carex pendula seedlings appearing like a green carpet after silt has been deposited after winter floods along the river terrace. They have dug out adult plants and weeded large areas of seedlings (Alice Shanks, pers.comm). It also grows in the shade under evergreen trees at Otahuna and Ernle Clark Reserve.

As well as colonising the banks of the Heathcote River, it has shown its potential to spread into other waterways and wetlands. If left uncontrolled it is likely to spread into the important wetlands and waterways of Canterbury such as Travis Swamp and Waihora Lake Ellesmere. The large infestation under trees at Otahuna is also good habitat for pigs that are known the live in that locality. Pigs could potentially spread Carex pendula seed onto the Port Hills where it could infest the tussock grassland and regenerating native forest.

Risks [under Section 72(1)(c)] Carex pendula is considered to have economic and conservation impacts. • Economic impacts Carex pendula is likely to have costly economic impacts if it spreads further down river from current sites and if it spreads into other waterways. The potential cost would be much reduced if Carex pendula is eliminated from the current sites. There is also a risk that it will spread into pasture thereby decreasing the land available for pasture and impacting on agricultural production. • Conservation impacts Because of its large size and prolific seeding it can displace native species in

86 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

a range of habitats. It is a threat to natural areas and restoration projects, especially near waterways and wetlands. Because it is shade tolerant it can form dense swathes under forest thereby preventing regeneration of native species in the understorey.

Section 72(1)(c) The early stage of invasion is reflected in the high score in the weed risk assessment. Early intervention at this stage could prevent Carex pendula from spreading to new high value areas such as Waihora Lake Ellesmere, Travis Swamp and on to the Port Hills.

The serious adverse effects, as described here, provide justification for the inclusion of Carex pendula in the RPMS (Section 72(1)(c).

87 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Carex pendula

88 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

29. Barberry (Berberis glaucocarpa)

Description Barberry is an evergreen or semi-deciduous spiny shrub that grows up to 5m in height. It has yellow flowers and reddish-black berries with a whitish bloom. It is a long-lived tree that produces many seeds. It tolerates a wide range of environments including poor soils, salt, wind, temperature variations and wet or dry conditions. However it is only tolerant of a small amount of shade.

Barberry has been planted as hedges in many parts of the country and has spread out of control in many areas. Trees are often present around old homesteads and near plantation forests. It produces copious seed which remain viable for a long time. The seed is spread by birds over large distances. Barberry can be hard to kill due to cut stumps resprouting.

Barberry can invade disturbed forest and shrubland, short tussock grassland and bare stony ground. It is regarded as one of the least desirable exotic species on Banks Peninsula (Wilson, 1999).

Current and potential habitat invasion Barberry is widespread in Canterbury. It is locally common in shrubby gullies in the Waitaki Valley, on Banks Peninsula, scattered in bush areas near Oxford, and scattered through coastal North Canterbury. It is a significant weed in the Waipara Gorge (Miles Giller, pers.comm.).

Barberry is likely to continue invading short tussock grassland, shrublands and forest margins.

Risks [under Section 72(1)(c)] Barberry is considered to have economic and conservation impacts. • Economic impacts The cost of controlling barberry is high. It is currently too widespread in Canterbury to eradicate so it is probably more cost effective to try to prevent new areas of infestation rather than trying to control it everywhere. • Conservation impacts Scattered plants and occasionally dense stands can replace native species. However, it is intolerant of deep shade so it is only competitive on the margins of forest or in forest with a poor canopy. In open environments like tussock grassland it does compete with native species.

Section 72(1)(c) The long time that barberry has been established in New Zealand and the numerous large infestations in Canterbury is reflected in the relatively low score in the weed risk assessment. To prevent ongoing spread of barberry considerable resources will be needed. Control of barberry should be targeted to areas of high conservation value.

The serious adverse effects, as described here, provide justification for the inclusion of barberry in the RPMS (Section 72(1)(c)).

89 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Barberry (Berberis glaucocarpa)

D. Cultivation and perceptions 1 Present as: mass plantings (3), frequent smaller plantings (2), 3 to 0 2 infrequent small plantings (1), not planted (0) 2 No. nurseries selling species: > 3 ,< 3, 3 to 0 1 3 Is it a crop plant? 1 or 0 0 4 Does it have unpleasant features? 1 or 0 0 Public attitudes score (sum D1–3 minus D4) 3

90 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

30. Elm (Ulmus sp.)

Description Elms are very large trees growing up to 40m high. They don’t produce viable seed but sucker prolifically and have epicormic shoots. Many of the elms cultivated in New Zealand are hybrids and are vigorous fast growing trees often seen in public parks and gardens.

Although elms are regarded as naturalised in New Zealand their impact on conservation and agricultural values is not of great concern. The suckering nature of these very large trees appears to be mainly of nuisance value in urban areas. When elms are cut down they resprout from the cut trunk and suckers may continue to appear for more than 25 years after cutting and treatment (Sally Tripp pers.comm.). This is because the suckers appear to detach from the main root mass after they become individual trees and therefore treating suckers may not have any effect on the large underground root mass. Suckering elms are a problem in native restoration areas where they deprive the desirable plants of moisture, nutrients and light. Of value to landowners would be information and guidance on how to get rid of elms and suckers if so desired. A mechanism to prevent the sale of elms in certain areas would be desirable.

Current and potential habitat invasion Elms have established locally throughout New Zealand in waste places and riverbeds often spreading from established cultivated trees within urban areas. Elms are not recorded as naturalized on Banks Peninsula (Wilson, 1999) but regarded as naturalised in Canterbury (Mahon, 2007). However, the Department of Conservation do control elm suckers in Hay Reserve on Banks Peninsula (Ian Hankin, pers. comm.).

It is likely that elms will continue to sucker and spread into unwanted places in Canterbury. However, as they do not spread from seed it is likely that they will remain close to homesteads and population centres.

Risks [under Section 72(1)(c)] Like many large trees with a suckering habit, elms are probably more of a nuisance value than having serious impacts on the environment. Elm suckers are detrimental to native restoration projects near population centres but are unlikely to be a problem in mature or regenerating native bush.

Section 72(1)(c) The large size of elms is reflected in the medium score of the weed risk assessment. However, the adverse effects of elms are not considered serious enough to include in the RPMS under Section 72(1)(c). It would be of value to landowners to provide information on ways to eliminate these large trees and suckers and also to explore ways of preventing the sale of these trees especially in urban areas.

91 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Elm (Ulmus sp.)

Points Score A. Interactions 1 Volume of individual plant m3: 10, 100, 1000, 10 000 1 to 4 4 2 Totally pre-empts sites, or covers native species to form canopy 2 or 0 2 3 Growth appears faster than associated native species 1 or 0 1 4 Species persists: < 5 years, 5–20 years, > 20 years 1 to 3 3 Impact score (Sum A1–4) 10

C. Reproduction 1 Species cryptic and cannot be detected before it reproduces 1 or 0 0 2 Produces viable seed 2 or 0 0 3 Seed dispersed primarily by: small birds, wind, or water (2), 2 or 1 1 large birds or passive/accidental dispersal (1) 4 Minimum generation time < 3 years (2), > 3 years (1) 2 or 1 1 5 Persistent vegetative organs above or below ground, 2 or 0 2 or seed bank (> 1 year) 6 Juveniles common within 100 m parents 1 or 0 1 Spread score (sum B1–3 + C1–6) 10 Impacts x spread score 100

D. Cultivation and perceptions 1 Present as: mass plantings (3), frequent smaller plantings (2), 3 to 0 1 infrequent small plantings (1), not planted (0) 2 No. nurseries selling species: > 3 ,< 3, 3 to 0 3 3 Is it a crop plant? 1 or 0 0 4 Does it have unpleasant features? 1 or 0 0 Public attitudes score (sum D1–3 minus D4) 4

92 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

31. Vipers bugloss (Echium vulgare)

Description Vipers bugloss is a bristly annual or biennial herb which can grow to nearly 1m high. It has blue funnel-shaped flowers and small seeds that are easily dispersed. Vipers bugloss grows and matures quickly and like many annual/biennials produces many long-lived seeds. It tolerates poor soils, wind and dry conditions. It is not particularly palatable but stock will graze it if food is scarce. Vipers bugloss grows on roadsides, river beds and on pasture and open waste land in dry areas. It is sometimes so abundant that in early summer, when it is flowering, roadsides and hillsides may be a sea of blue.

Current and potential habitat invasion Vipers bugloss is widespread and common on disturbed sites throughout Canterbury especially inland dry places eg. Molesworth.

Vipers bugloss tends to invade dry disturbed land but as with most annual/biennials it is dependent on favourable conditions for good germination and growth and may vary in cover from year to year.

Risks [under Section 72 (1)(c)] Economic impacts Vipers bugloss has both positive and negative economic impacts. The flowers are a good source of nectar for honey and this is an economic benefit for beekeepers. Farmers tend to dislike the bristly hairy leaves and stems of vipers bugloss that stick to the sheeps wool. However vipers bugloss, being tolerant of dry conditions may provide a food source for stock when there is little else. Conservation impacts Vipers bugloss may have a temporary smothering effect in open, disturbed, low growing indigenous plant communities. However as it is an annual or biennial this effect is temporary and dependent on suitable conditions for good seed germination. Mostly vipers bugloss grows on disturbed ground such as roadsides, riverbeds and grazed dry low fertility sites where there are few native plant communities. If vipers bugloss is grubbed or sprayed the ground becomes bare and disturbed which will encourage other similar weeds to take its place, so little is gained. Control may be prudent if there are natural areas of high value where vipers bugloss is not yet present eg. the Heron Basin in inland Canterbury. At Kaitorete Spit, controlling vipers bugloss at Birdlings Flat may prevent it invading the sand dunes where there are several threatened plant communities. Probably the most effective method of reducing the amount of vipers bugloss is to prevent disturbance thereby discouraging seed germination. Where vipers bugloss is unwanted and present in grazed situations then removal of grazing may be enough to reduce vipers bugloss to minor amounts. Naturally disturbed situations like riverbeds may require control by spraying or grubbing in areas of high conservation value.

Section 72 (1)(c) Although vipers bugloss has some detrimental effects on conservation and agricultural values they are not considered serious enough to warrant inclusion in the RPMS. However vipers bugloss could be considered a ‘site led’ weed where it could be controlled to prevent it spreading into areas of high conservation value eg. Kaitorete Spit and the Heron Basin.

93 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

Vipers bugloss (Echium vulgare)

Points Score A. Interactions 1 Volume of individual plant m3: 10, 100, 1000, 10 000 1 to 4 1 2 Totally pre-empts sites, or covers native species to form canopy 2 or 0 0 3 Growth appears faster than associated native species 1 or 0 1 4 Species persists: < 5 years, 5–20 years, > 20 years 1 to 3 1 Impact score (Sum A1–4) 3

C. Reproduction 1 Species cryptic and cannot be detected before it reproduces 1 or 0 1 2 Produces viable seed 2 or 0 1 3 Seed dispersed primarily by: small birds, wind, or water (2), 2 or 1 2 large birds or passive/accidental dispersal (1) 4 Minimum generation time < 3 years (2), > 3 years (1) 2 or 1 2 5 Persistent vegetative organs above or below ground, 2 or 0 2 or seed bank (> 1 year) 6 Juveniles common within 100 m parents 1 or 0 1 Spread score (sum B1–3 + C1–6) 9 Impacts x spread score 27

94 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

32. (Muehlenbeckia australis)

Description Muehlenbeckia is a much branched perennial climbing vine that can climb up to 10m high. Flowers are in panicles and seeds are small and enclosed in fleshy cups. Muehlenbeckia is one of the few native species that can be considered to have some weedy characteristics. It can grow over scrub and forest margins reducing light to regenerating shrubs and trees and thereby slowing down the successional process. However the native muehlenbeckia has redeeming qualities not present in exotic vines. Although both native and introduced climbers can overwhelm other plants the introduced vine is usually much more aggressive. As mature vines, old mans beard and Chilean flame-creeper, can engulf the boles of large native trees in a mass of stems whereas the native muehlenbeckia has an ascending cable with a small leafy head up in the canopy (Godley, 2006). Old mans beard smothers the understorey whereas muehlenbeckia is not shade tolerant. Muehlenbeckia is not seen in mature forest as seeds won’t germinate in deep shade. Muehlenbeckia also has wider ecosystem benefits not seen in most exotic vines. There is an incredibly rich moth and butterfly fauna associated with muehlenbeckia with the native copper butterfly caterpillars feeding on it at the forest margins. The flowers and berries provide food for lizards, invertebrates and native aphids. Old mans beard provides little food and is spread far and wide as the fluffy seeds are carried in the wind whereas Muehlenbeckia seed is not wind dispersed meaning it is not spread over great distances.

Current and potential habitat invasion In Canterbury muehlenbeckia grows in regenerating scrub and forest margins on Banks Peninsula, coastal areas and inland.

It is likely that muehlenbeckia distribution will remain similar to the current distribution unless there are big disturbance events.

Risks [under Section 72(1)(c)] Muehlenbeckia is not considered to have serious impacts on section 72(1)(c) matters.

Section 72(1)(c) Although muehlenbeckia is capable of forming a canopy over regenerating forest and slowing down regeneration it has compensating qualities of providing food and habitat for native lizards, invertebrates, moths and butterflies. It is not considered an agricultural weed as it does not occupy grassland habitats preferring scrub and forest margins. Although this vine does have impacts they are considered minor and therefore it does not warrant inclusion in the RPMS (Section 72(1)(c).

95 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Weed Risk Assessment

(Muehlenbeckia australis)

Points Score A. Interactions 1 Volume of individual plant m3: 10, 100, 1000, 10 000 1 to 4 2 2 Totally pre-empts sites, or covers native species to form canopy 2 or 0 2 3 Growth appears faster than associated native species 1 or 0 0 4 Species persists: < 5 years, 5–20 years, > 20 years 1 to 3 2 Impact score (Sum A1–4) 6

96 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

33. Townsville stilo (Lotus sp.)

Townsville Stilo is Stylosanthes humilis and not a Lotus species. Townsville stilo is endemic to Central and South America and occurs in tropical areas where it is one of the most important pasture legumes. It is naturalised in Northern Australia but is not known from Canterbury or New Zealand (Mahon, 2007) (Howell and Sawyer, 2006).

As the species name is not known then it cannot be considered for inclusion on the RPMS.

97 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

98 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

3.4 Acknowledgements Thanks to Christchurch City Council, Department of Conservation, Environment Canterbury, QEII staff and individuals who provided information for this report, in particular, Gemma Bradford, Helen Braithwaite, Di Carter, Miles Giller, Ian Hankin, Alan McDonald, Ray Maw, Rob McCaw, Trevor Partridge, Alice Shanks, Sally Trip and Manfred von Tippelskirch.

3,5 References Champion, Paul D. 2005. Evaluation criteria for assessment of candidate species for inclusion in the National Pest Plant Accord. Report prepared for Biosecurity New Zealand. NIWA Project: MAF05204 Connor, H.E. 1977. The poisonous plants in New Zealand. DSIR. Christchurch. Craw, C.J. 2000. Weed Manager. A guide to the identification, impacts and management of conservation weeds of New Zealand. Department of Conservation Edgar, E; Connor,H.E. 2000. Flora of New Zealand. Volume V. Gramineae. Manaaki Whenua Press, Lincoln. New Zealand. Gatehouse, Hazel. 2009. Literature Review of Carex pendula. Godley, Eric. 2006. A Botanist’s Notebook. Manuka Press, Christchurch. Harris, S.; Timmins, S.M. 2009: Estimating the benefit of early control of all newly naturalised plants. Science for Conservation 292. Department of Conservation, Wellington. 25 p. Healy, A.J; Edgar, E. 1980. Flora of New Zealand. Vol.III. Adventive Cyperaceous, Petalous and Spathaceous Monocotyledons. DSIR, Wellington. Heenan et al. 1999. Checklist of dicotyledons, gymnosperms, and pteridophytes naturalised or casual in New Zealand: additional records 1997-1998. New Zealand Journal of Botany 37. Heenan et al. 2009. Additional records of indigenous and naturalised plants with observations of Gunnera tinctoria, on Stewart Island, New Zealand. New Zealand Journal of Botany 47. Howell, Clayson. 2008. Consolidated list of environmental weeds in New Zealand. DOC Research & Development Series 292. Department of Conservation, Wellington. Howell, Clayson; Sawyer, John. 2006. New Zealand Naturalised Vascular Plant Checklist. New Zealand Plant Conservation Network. Lovis, J. 1980. A puzzling Polypodium on the Port Hills. Canterbury Botanical Society Journal 14: 55-57. Mahon, D.J. 2007: Canterbury naturalised vascular plant checklist. Department of Conservation, Christchurch. Roy, B; Popay, I; Champion, P; James, T; Rahman,A.1998. An illustrated guide to Common Weeds of New Zealand. New Zealand Plant Protection Society. Sykes, W.R; Williams, P.A. 1999. New Zealand Botanical Society Newsletter 55. Van der Mast, S. & J. Hobbs.1998. Ferns for New Zealand Gardens. Godwit Publishing Ltd., Auckland. Williams, P.A.; Newfield, M. 2002: A weed risk assessment system for new conservation weeds in New Zealand. Science for Conservation 209. 23 p. Williams, P.A.; Boow, J.; La Cock, G.; Wilson, G. 2005: Testing the weed risk assessment system for new conservation weeds in New Zealand. DOC Research & Development Series 225. Department of Conservation, Wellington. 19 p. Webb, C.J.; Sykes, W.R.; Garnock-Jones, P.J. 1988. Flora of New Zealand. Volume IV. Naturalised Pteridophytes, Gymnosperms, Dicotyledons. DSIR, Christchurch, New Zealand. Webb, C.J; Sykes, W.R; Garnock-Jones; Brownsey, P.J. 1995. Checklist of dicotyledons, gymnosperms, and pteridophytes naturalised or casual in New Zealand: additional records 1988-1993. New Zealand Journal of Botany 33. Wilson, Hugh D. 1999. Naturalised Vascular plants of Banks Peninsula. A Canterbury Botanical Society Special Publication.

99 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

3.6 Other sources of information: Weed distribution maps of Canterbury from an unpublished Department of Conservation project ‘Presence/absence of plant pests in New Zealand’. National Pest Plant Accord Manual. 2008. Technical Advisory Group Assessment of National Pest Plant Accord Species Biosecurity website: http://biosecurity.govt.nz Biosecurity New Zealand. (2001). National Plant Pest Accord http://www.biosecurity.govt.nz/pests-diseases/plants/accord.htm Canterbury Weed Guide. 2002. A Selection of weeds from Christchurch and Canterbury. NZ Plant Conservation Network website: www.nzpcn.org.nz Weedbusters website: http://weedbusters.co.nz 'The New Zealand Plant Finder Database available at www.plantfinder.co.nz '

100 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Appendix 1: Weed Risk Assessment (total scores for each species) Note: Weed risk assessments from Williams, 2005 Note: The total score gives a ranking in order of priority for control * National Plant Pest Accord (NPPA) species Species Impact Spread Combined risk Public attitudes Total Rank (impact x spread) (combined risk + public attitudes) * False tamarisk (Myricaria germanica) 4 24 168 0 168 1 Puna grass (Achnatherum caudatum) 7 23 161 0 161 2 * Senegal tea (Gymnocoronis spilanthoides) 7 22 154 2 156 3 * Horsetail (Equisetum hyemale) 7 22 154 1 155 4 * Bomarea (Bomarea caldasii) 9 17 153 1 154 5 * Chilean flame creeper (Tropaeolum speciosum) 9 17 153 1 154 6 * Yellow waterlily (Nuphar lutea) 7 22 154 0 154 7 * Japanese spindle tree (Euonymus japonicus) 7 20 140 1 141 8 * Climbing asparagus (Asparagus scandens) 8 17 136 0 136 9 * Chilean rhubarb (Gunnera tinctoria) 7 19 133 2 135 10 Common polypodium (Polypodium vulgare ) 7 18 126 1 127 11 Carex pendula 7 18 126 1 127 12 * Giant hogweed (Heracleum mantegazzianum) 7 18 126 0 126 13 * Royal fern (Osmunda regalis) 7 18 126 0 126 14 Russell lupin (Lupinus polyphyllus) 7 17 119 5 124 15 * Purple loosestrife (Lythrum salicaria) 7 17 119 1 120 16 * Smilax (Asparagus asparagoides) 7 17 119 0 119 17 * Pigs ear (Cotyledon orbiculata) 7 16 112 1 113 18 * Cotoneaster simonsii 7 16 112 1 113 19 * Asiatic knotweed (Reynoutria japonica) 7 16 112 0 112 20 * Moth plant (Araujia sericifera) 9 12 108 2 110 21 * African club moss (Selaginella kraussiana) 7 15 105 0 105 22 Elm (Ulmus sp.) 10 10 100 4 104 23 * Giant knotweed (Reynoutria sachalinensis) 7 14 98 1 99 24 * Yellow flag iris (Iris pseudacorus) 7 14 98 0 98 25 * Madeira vine (Anredera cordifolia) 9 12 96 1 97 26 * Green goddess (Zantedeschia spp) 7 13 91 1 92 27 Boxthorn (Lycium ferocissimum) 8 11 88 0 88 28 * Grey willow (Salix cinerea) 89 720 72 29 Barberry (Berberis glaucocarpa) 79 633 66 30 Muehlenbeckia australis 68 480 48 31 101 Vipers bugloss (Echium vulgare) 39 270 27 32 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

102 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

Part 4: Animals for possible inclusion in a RPMS

4.1 European hedgehog (Erinaceus europaeus)

Hedgehogs are small, spiny, mainly insectivorous nocturnal animals, most closely related to shrews and moles. They have the ability to roll into a tight prickly ball for defensive purposes.

Hedgehogs are mostly abundant throughout lowland districts, less numerous in the hills and rare in mountainous areas. They are abundant on temperate lowland and farmland where frosts are few and mild, and where food is abundant. Lowland stream and river sides are also favoured habitats. Cities and suburbs also support dense populations of hedgehogs, because invertebrates and dry sites for hibernating are available, as well as extra food purposely provided by householders.

Hedgehogs are mainly insectivorous, but will eat any animal substance and even some plant material. Hedgehogs may eat 160 g of invertebrates per animal per day. Diets vary depending on site and season, but beetles are important foods in most habitats. In suburban areas and lowland farms, hedgehogs eat mainly slugs, snails and a great variety of ground insects and larvae. Earthworms are commonly eaten in pasture, but rarely in forest or drylands where weta and grasshoppers are more important. Earwigs and Lepidopteran larvae are eaten in large numbers where available. Hedgehogs also feed on mice, lizards, frogs, eggs and chicks of ground-nesting birds, and scavenge carrion.

The effects of hedgehogs on indigenous fauna in New Zealand have not been quantified although they clearly have the potential to contribute significantly to the decline of numerous taxa, including threatened ground-nesting birds. For example, hedgehogs were responsible for two of every three losses of New Zealand dotterel nests among sand dunes at Tawharanui. Competition from hedgehogs could limit kiwi numbers in the long-term because kiwi and hedgehogs have similar diets and nest in similar sites.

C.M. King (Ed) 2005: The Handbook of New Zealand Mammals, Second Edition, Oxford University Press, Melbourne.

103 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

4.2 Ship rat (Rattus rattus), Norway rat (Rattus norvegicus) There are two introduced European rat species in New Zealand. The ship rat has a pointed muzzle, large eyes and ears. The tail is longer than the combined length of the head and body. The body is quite sleek, with a scaly, sparsely haired tail. Ship rats are smaller than Norway rats, weighing 130–170 g. The Norway rat is the largest rat in New Zealand (often weighing 150–300 g), but can grow to more than 500g. It has a short body and a heavy tail, which is slightly shorter than the combined length of the head and body. The coat of both sexes is coarse and quite shaggy, greyish brown on the flanks with a darker brown along the back. The stomach and throat are pale grey. Norway rats are competent swimmers and are commonly called ‘water rats’. This ability enables them to colonise offshore islands. In favourable conditions a crossing of 600m is possible. They can also jump up to 77cm vertically or 120cm horizontally.

Rats are notorious vectors for the spreading of human diseases. Rats have been responsible for the extinction of a number of native species1 and they continue to have a major impact on New Zealand’s flora and fauna. They consume seeds and foliage, birds, eggs, invertebrates, snails and lizards. Ship rats eat seeds, fruits, flowers and other plant parts, which make up 80 per cent (by volume) of their diet. The damage they cause is difficult to separate out from the damage caused by the suite of other rodents and herbivores also occupying their range. Norway rats tend to occupy coastal margins, but are also found in forests.

In mixed podocarp-hardwood forest a common sign of ship rats is the cached and gnawed remnants of miro (Prumnopitys ferruginea) or hinau (Elaeocarpus dentatus) seeds. Although they destroy many seeds, ship rats may also help to disperse some seeds, as shown in captive feeding trials. On the mainland, historical damage to fauna by ship rats is difficult to distinguish from the damage from kiore, Norway rat and mustelid invasions that preceded them2.

Ship rats are found from sea level to tree line, and in a broad range of habitats, including urban areas, farmland, both native and exotic forests, and shrubland. They are nocturnal, excellent climbers and are probably the most widespread mammal predator found in non- beech forests on the New Zealand mainland. They reach their highest densities in lowland podocarp-broadleaved forests. Densities have been recorded that show there are 1.7 rats per ha in Orongorongo Valley (measured over 29 months), 2.9 rats per ha at Puketi before spring breeding, and 6.2 rats per ha in summer at Rotoehu.

Insects including beetles, moths, stick insects, cicadas and especially weta, are always eaten when available. Only in New Zealand is there a seasonal predominance of arthropods in the diet. In areas where rat control has taken place, increases in insect abundance have been observed.

C.M. King (Ed) 2005: The Handbook of New Zealand Mammals, Second Edition, Oxford University Press, Melbourne.

104 Section 72 Analysis for 5-year Review Results Canterbury Regional Pest Management Strategy 2005-2015

4.3 Argentine Ant

105 Potential impact of the Argentine ant (Linepithema humile) in New Zealand and options for its control

SCIENCE FOR CONSERVATION 196

R.J. Harris

Published by Department of Conservation P.O. Box 10-420 Wellington, New Zealand Science for Conservation is a scientific monograph series presenting research funded by New Zealand Department of Conservation (DOC). Manuscripts are internally and externally peer-reviewed; resulting publications are considered part of the formal international scientific literature. Titles are listed in the DOC Science Publishing catalogue on the departmental website http:// www.doc.govt.nz and printed copies can be purchased from [email protected]

ISSN 11732946 ISBN 0478222459

This report was prepared for publication by DOC Science Publishing, Science & Research Unit; editing and layout by Ian Mackenzie. Publication was approved by the Manager, Science & Research Unit, Science Technology and Information Services, Department of Conservation, Wellington. CONTENTS

Abstract 5

1. Introduction 6

2. Objectives 6

3. Biology 6 3.1 Tramp status 6 3.1.1 Strong tendency to move and associate with humans 7 3.1.2 Unicoloniality 7 3.1.3 Interspecific aggression 7 3.1.4 Polygyny 7 3.1.5 Mating and budding 7 3.2 Diet 8 3.3 Dispersal 8 3.4 Abundance 8 3.5 Distribution 9 3.5.1 Climate associations 9 3.5.2 Habitat associations 10

4. Impacts 11 4.1 Impacts on native systems 11 4.2 Human and economic impacts 13

5. Control options for L. humile 14 5.1 Chemical control 14 5.2 Biological control 16

6. L. humile in New Zealand 17 6.1 New Zealands ant fauna 17 6.2 L. humile arrival and spread 17 6.3 Potential future distribution in New Zealand 19 6.3.1 Methods 19 6.3.2 Results 21

7. Tiritiri Matangi Island eradication trial 24 7.1 Background 24 7.2 Methods 25 7.2.1 Poison baiting 25 7.2.2 Ant monitoring 26 7.3 Preliminary results and discussion 26 7.3.1 Bait application 26 7.3.2 Bait effectiveness 26 7.3.3 Evaluation of non-toxic baits for monitoring 27 7.3.4 Other ants 28 7.3.5 Overview of strategy and future plans 29 8. Conclusions 30

9. Recommendations 31

10. Acknowledgements 31

11. References 32

4 HarrisPotential impact of the Argentine ant in New Zealand Potential impact of the Argentine ant (Linepithema humile) in New Zealand and options for its control

R.J. Harris Landcare Research, Private Bag 6, Nelson, New Zealand

ABSTRACT

The Argentine ant (Linepithema humile) is a highly invasive tramp species from South America that has spread to many countries of the world. Human-assisted transportation of ant colonies is the main mechanism of dispersal to new areas. L. humile was first found in Auckland in 1990. It is now widespread within Auckland, but still patchily distributed, and has established at a number of other sites around New Zealand from Northland to Canterbury. Although mainly established in urban areas, the first population reported from conservation land administered by the Department of Conservation has established on Tiritiri Matangi Island. The potential distribution of L. humile is predicted using temperature and land cover data. Much of Northland and coastal areas of the North Island are considered to be at risk of establishment. Except for urban areas, most of central and southern New Zealand is considered too cold. Indigenous scrub/low stature vegetation is the native habitat considered most at risk, while indigenous forest is unlikely to be invaded. Many offshore islands contain suitable habitat for the establishment of L. humile should it be transported there. Spread to areas outside urban development will be slow, as unassisted dispersal is only about 150 m/yr. At sites of establishment, most other ant species will be replaced by L. humile, and the structure of the non-ant invertebrate community will be altered. Currently, baiting is considered the most effective method for control, and several bait products are commercially available. For areas of high conservation significance, such as offshore islands, eradication of populations is desirable rather than management requiring ongoing chemical application. An eradication strategy is being tested on Tiritiri Matangi Island and two urban populations using the insecticide fipronil in an experimental protein bait developed in Australia. The first treatment has resulted in a huge reduction in the density of L. humile, but follow-up treatment is required to achieve eradication. If successful, the strategy will provide a method to eliminate L. humile from important conservation areas. Eradication is labour intensive so is best undertaken while infestations are small, requiring targeted monitoring of key conservation sites. Keywords: Argentine ant, Linepithema humile, New Zealand, review, impact, control, predicted distribution

© May 2002, Department of Conservation. This paper may be cited as: Harris, R.J. 2002. Potential impact of the Argentine ant (Linepithema humile) in New Zealand and options for its control. Science for Conservation 196. 36 p.

Science for Conservation 196 5 1. Introduction

The Argentine ant (Linepithema humile (Mayr)) (Family Formicidae, Subfamily Dolichoderinae) is a highly invasive South American species that has spread to many countries. The first report of L. humile establishment in New Zealand was in 1990, when a small population was found in Auckland (Green 1990). The initial aim of this research project, initiated by Dr Chris Green, Auckland Conservancy, Department of Conservation was to review the threat L. humile posed to New Zealand native ecosystems and what options exist to control it should it establish in highly valued native ecosystems. However, soon after undertaking this project, a population of L. humile was discovered on Tiritiri Matangi Island, an island of high conservation value in the Hauraki Gulf. As a consequence, the need for an eradication strategy became imperative, and a trial was initiated on Tiritiri Matangi. This report reviews published information on the biology, impacts, and environmental requirements of L. humile, and relates findings to the likely impact of this species in New Zealand. The management options available to control the ant are reviewed, and the results of the Tiritiri Matangi eradication trial are reported.

2. Objectives

The objective of this research was to evaluate the potential impact of L. humile in New Zealand and identify options for their management by: Reviewing international scientific literature on their biology, impacts, and control methods Collecting information on the distribution of L. humile in New Zealand Using current knowledge on habitat preferences and temperature limitations to predict the potential distribution of L. humile in New Zealand Detailing the eradication trial in progress on Tiritiri Matangi Island

3. Biology

3.1 TRAMP STATUS

L. humile (Fig. 1) is commonly referred to as one of the tramp ant species (Passera 1993). Other tramp species include the big-headed ant (Pheidole megacephala Fab.) and the red imported fire ant (Solenopsis invicta Buren). Tramp species generally have the following features in common that have enabled them to be highly successful invaders.

6 HarrisPotential impact of the Argentine ant in New Zealand Figure 1. A worker of the Argentine ant, Linepithema humile.

3.1.1 Strong tendency to move and associate with humans The species are able to relocate and survive in response to high levels of disturbance. L. humile workers pick up larvae and eggs and relocate their nest in response to changes in weather, human activity, and/or food supply. Their close association with human activity means that human-assisted transportation (e.g. in potted plants, rubbish, freight) is the main mechanism of long distance jump dispersal (Suarez et al. 2001).

3.1.2 Unicoloniality In their native ranges, populations of L. humile exhibit pronounced inter- colony aggression (Tsutsui et al. 2000). However, adventive populations of L. humile are not aggressive to conspecifics from different nests, effectively operating as one large colony (unicoloniality). This allows them to support higher densities and maintain greater foraging activity (Holway et al. 1998). The lack of inter-colony competition may well be due to the reduced genetic diversity of introduced populations (Chen & Nonacs 2000; Liang & Silverman 2000).

3.1.3 Interspecific aggression L. humile is highly effective at recruiting foragers to, and monopolising, food resources. Workers are highly aggressive to other ant species, and although individual workers frequently lose these aggressive interactions, their numerical dominance means that L. humile succeeds in displacing other ants (Holway 1999).

3.1.4 Polygyny Colonies have several queens, resulting in a high reproductive rate. L. humile has 0.1 to 1.6 queens per 100 workers (Keller et al. 1989).

3.1.5 Mating and budding L. humile, and tramp species generally, mate in the nest, with few or no nuptial flights by queens, and the colonies disperse by budding (a group of workers and

Science for Conservation 196 7 queens separating from the main colony and walking to a new site). Jump- dispersal only occurs passively (e.g. via flooding or human-assisted transportation). Polygynous colonies of S. invicta disperse primarily by budding, while new queens from single queened colonies undergo nuptial flights and some also disperse by budding (Vinson & Sorensen 1986). Other features common to tramp species include their small size, monomorphic workers (except P. megacephala), a short queen lifespan (effectively <1 year in L. humile; Keller & Passera 1990), and worker sterility.

3.2 DIET

L. humile has a generalist diet in native systems that includes nectar, insects, seeds, carrion, and honeydew secreted by homopterans (Suarez et al. 1998). Diet overlap with other ant species is high. The food carried by L. humile foragers is predominately liquid (distended gaster91.6%), with live or dead insects (5.5%), unidentified loads (2.5%) and seeds (0.5%) making up the remainder (Human et al. 1998). The diet of workers consists primarily of sugar, while larvae and queens are mainly fed protein (Vega & Rust 2001). Queens of L. humile are unusual in that they take part in foraging and grooming (Vega & Rust 2001).

3.3 DISPERSAL

L. humile often has a very patchy distribution due to the interaction of its two modes of dispersal, diffusion and jump-dispersal (Holway 1995; Barber 1916 in Suarez et al. 2001). The rate of spread due to budding is relatively slow, averaging about 150 m/yr where climate or habitat is not limiting (reviewed by Suarez et al. 2001). However, the rate can range from near zero in areas of climatic extremes up to 800 m/yr in highly favourable recently invaded habitat (Holway 1998a; Way et al. 1997). Queens appear to need workers to establish a colony successfully, but a colony with as few as 10 workers can grow quickly in the laboratory (Hee et al. 2000). In the last century, human-assisted dispersal has resulted in the successful establishment of L. humile on six continents and many oceanic islands (Suarez et al. 2001). Establishment patterns in many areas follow human transport networks (roadways and towns) (e.g. Holway 1995; Human et al. 1998). Assisted dispersal is via transport of goods that contain nests, including soil, pot plants, foodstuffs, vehicles, appliances, and garbage (Van Schagen et al. 1993; Suarez et al. 2001).

3.4 ABUNDANCE

L. humile is usually extremely abundant wherever it has established outside its native range. Any suitable nesting site will be occupied, and thousands of ants will pour out if the nest is disturbed. Nests are inter-connected via foraging

8 HarrisPotential impact of the Argentine ant in New Zealand trails, and workers are freely exchanged between them. These features prevent accurate density estimates and hence no estimates of absolute density appear in the literature. The activity along foraging trails indicates the high abundance. Markin 1967 (in Vega & Rust 2001) estimated 50,000600,000 ants ascending single trees daily. Baits cannot be used to assess absolute density or species composition at a site, as L. humile recruit to and monopolise baits (Suarez et al. 1998). Unbaited pitfalls traps catch few L. humile (J. Haw pers. comm.) and differences in methodologies where baited pitfalls have been used make comparison of abundance invalid. In general, studies using baited pitfall traps tended to catch greater numbers of ants in pitfalls after invasion by L. humile. But care is needed in this interpretation, as it may reflect activity patterns not abundance. Not all studies showed that total ant biomass was increased after L. humile invasion (e.g. Holway 1998b), but no increase was observed in areas with a species-rich ant fauna before invasion (e.g. Holway 1998b). There are considerable seasonal variations in the density and distribution patterns of L. humile populations. In favourable conditions, new nests containing queens and workers bud off and the infestation expands, whereas, in adverse climatic conditions, small nests merge to form fewer larger colonies (Davis & Van Schagen 1993). Although L. humile frequently displaces other ant species, it is itself occasionally displaced. In parts of the USA, L. humile has been displaced by the invasion of the red imported fire ant, S. invicta (Porter et al. 1988). P. megacephala appears to survive in the presence of L. humile, but not coexist there is a dynamic equilibrium between the two species, with each species holding possession of sites for extended periods (Haskins & Haskins 1988).

3.5 DISTRIBUTION

3.5.1 Climate associations

L. humile mostly occurs in 3036o latitude belts of both hemispheres (Majer 1993). Isolated populations can establish and thrive outside these latitudes, where climatic conditions are favourablee.g. California, 3738.5o (Holway 1995; Ward 1987); Hawaii, 20.521o (Cole et al. 1992); Easter Island, 27o (Morrison 1997). In Hawaii, coastal areas appear to be too hot for L. humile, and it is restricted to higher, cooler elevations from 2070 m to 2880 m with mean monthly temperatures ranging from 9.8oC to 13.5oC (Cole et al. 1992). During cooler times of the year foraging will be restricted, as 10oC is the reported lower limit for foraging (Markin 1970). Foragers are most active when the temperature ranges from 1030oC, and cease foraging when surface temperatures reach 32oC or drop below 15oC (Hedges 1998). In arid climates, a lack of water probably restricts the distribution of L. humile (Ward 1987; Van Schagen et al. 1993; Kennedy 1998). In the colder states of the US, such as Minnesota and Illinois, L. humile cannot survive outside human modified landscapes (Suarez et al. 2001) where human activities and structures create warmer microclimates. These states have extremely cold winters and hot summers (e.g. Duluth, Minnesota, has an average temperature of 13.6oC in the coldest month and 17.5oC in the hottest).

Science for Conservation 196 9 Within a site, distribution can be patchy and related to microclimate. For example, in a citrus orchard, 93% of colonies were found in the south-west (hotter) quadrant around orchard trees (Phillips 1986). Oviposition slows over the winter and development rates of eggs, larvae, and pupae slow. Oviposition does not occur below a daily mean temperature of 18.3oC, and the population reaches a seasonal low in numbers during winter (Vega & Rust 2001). Environments with high rainfall reduce foraging time and may reduce establishment chances (Vega & Rust 2001).

3.5.2 Habitat associations Although frequently associated with human settlement, L. humile is not restricted to modified habitats and is present in native vegetation in a number of locations (e.g. HawaiiReimer 1993, CaliforniaWay et al. 1997, South Af- ricaGiliomee 1986). In Hawaii, L. humile is present in undisturbed montane habitat in dry and mesic areas, and shrubland/grassland sites with from 1015% to nearly 100% vegetation cover, but is not present in wet forests (Reimer 1993; Cole et al. 1992). In San Diego County, Southern California, L. humile is present in scrub habitat fragments, and along the edge of a large continuous area of native vegetation (Suarez et al. 1998). However, 100 m into the fragments, L. humile numbers begin to decline. Conversely, native ants were only in habitat >200 m from an urban edge. Many of the smaller fragments did not have a central pocket of native ants. It is unclear if L. humile was still invading the fragments or not penetrating further into the scrub fragments because they lacked water (Suarez et al. 1998). Information is scarce on whether substrate influences the distribution of L. humile. In western and southern Portugal, it is common in sand and clay loam soils, but is largely absent in sandy loam soils associated with metamorphic rocks, despite otherwise favourable conditions (Way et al. 1997). The distribution of L. humile in Portugal has not changed much in the last 40 years, and Way et al. (1997) conclude this is not due to dispersal limitations. World-wide, the majority of reports of L. humile are from urban areas (Suarez et al. 2001). Where L. humile has invaded native ecosystems it has predominantly been low stature scrub vegetation (Giliomee 1986; Reimer 1993; Way et al. 1997; Human et al. 1998). Forests have either not been invaded (Reimer 1993; Cole et al. 1992) or the forest margins alone have been inhabited (Suarez et al. 1998). L. humile is also reported to be a pest of horticultural land (Davis & Van Schagen 1993).

10 HarrisPotential impact of the Argentine ant in New Zealand 4. Impacts

4.1 IMPACTS ON NATIVE SYSTEMS

A wide range of L. humile impacts on native systems have been documented (Table 1). L. humile frequently displaces most ant species (e.g. Ward 1987; De Kock 1990; Cammell et al. 1996; Human & Gordon 1996; Suarez et al. 1998; Sanders et al. 2001). Changes in the ant community composition can have flow on effects on the ecosystem as ants not only constitute a large component of the total animal biomass, but can also act as engineers affecting soil processes (Folgarait 1998). Overseas, L. humile is consistently better than the native ants at exploiting food resources in terms of speed of locating food, recruiting large numbers of workers to the food, and the length of foraging period (Human & Gordon 1996). Both direct interference and exploitation of food resources appear important in displacement (Human & Gordon 1996; Holway 1999). L. humile overruns nests of larger species through physical aggression and numerical dominance, with displacement occurring within several hours (De Kock 1990). Epigaeic (above ground foraging) species are more affected than hypogaeic (below ground foraging). Native ant species will rapidly re-establish following removal of L. humile through baiting (P. Davis pers. comm.). One epigaeic species that has been documented to survive in the presence of L. humile is most active in winter months (Suarez et al. 1998), another produces a defence secretion that keeps L. humile away from feeding sites (De Kock 1990). Those that do remain in the presence of L. humile are often cryptic rarities that are seldom encountered (Haskins & Haskins 1988). There are several cryptic hypogaeic species in New Zealand (all adventive) that potentially could co-exist, but these are forest species and consequently likely

TABLE 1. SUMMARY OF THE MAIN DOCUMENTED IMPACTS OF L. humile in NATIVE SYSTEMS.

MODIFICATION IMPACT ON MECHANISM SELECTED REFERENCES

1. Community structure Ant diversity Interference competition and food Human & Gordon 1996; monopolisation Holway 1999 Abundance and diversity Interference and resource competition; Cole et al. 1992; Way et al. of other invertebrates predation on eggs, larvae and adults 1992; Human & Gordon 1997 Abundance of vertebrates Interference and resource competition Suarez et al. 2000

2. Community processes Pollination Competition for nectar with effective Buys 1987; pollinators Visser et al. 1996 Seed dispersal/ Displacement of specialist ants that have Bond & Slingsby 1984; Regeneration co-evolved to assist seed dispersal and Giliomee 1986 seedling germination Decomposition/ Changing the guild structure of the Ward 1987; De Kock 1990; nutrient cycling invertebrate community Folgarait 1998

Science for Conservation 196 11 to have little overlap with L. humile. It is likely that the adventive species, Cardiocondyla minutior, newly established in Mt Maunganui (Harris & Berry 2001), will coexist with L. humile. No native ants are at risk of extinction from L. humile spread, because of the wide distribution of native ant species and their occurrence in indigenous forest. The strong competitive ability of L. humile, together with its broad diet, mean that through direct predation (Human & Gordon 1997), competition, interference, and egg predation (e.g. cerambycidsWay et al. 1992), it will interact with many invertebrate species in any habitat where it establishes colonies. The conclusion from impact studies is that when the total ant biomass is increased following the invasion of L. humile, the invertebrate community is negatively impacted (Human & Gordon 1997; Cole et al. 1992). Alternatively, if total ant biomass after invasion is similar to pre-invasion levels, the abundance and diversity of invertebrates (other than ants) is similar across the invasion front (Holway 1998b). A limitation of all these studies is that they have not experimentally reduced L. humile behind an invasion front and looked at community recovery to demonstrate conclusively that differences across the invasion front are due to L. humile. Detrimental impacts of L. humile invasion are documented from sites in Hawaii and California (Cole et al. 1992; Human & Gordon 1997; Bolger et al. 2000). For all these studies the invertebrate fauna was more diverse in the absence of L. humile and the non-ant invertebrate abundance was higher. Many groups of invertebrates were entirely absent or rare in the presence of L. humile (e.g. muscid flies, springtails, cunipid wasps, ticks, or mites). Some taxa, mostly scavengers, were relatively more abundant in the invaded areas. Many of these were adventive species, such as some carabid beetles, non-native Isopods, Dermaptera, and Blattaria. In New Zealand, the total ant biomass is likely to increase with the establishment of L. humile, and community assemblages will be modified. Many endemic species will be adversely affected and localised extinctions are likely, placing species with restricted distribution at risk. Particularly at risk are those that occur in coastal/scrub vegetation in northern New Zealand, for example, flax snails Placostylus ambagiosus (Sherley 1996). Generally, ants are considered poor pollinators, so a reduction in other pollinators is likely to be detrimental to flowering plants. Buys (1987) demonstrated that L. humile collected large amounts of nectar from Eucalyptus spp. before native bees began foraging, Visser et al. (1996) found that the abundance and richness of pollinators on Protea were reduced when there were greater than 200 L. humile per infested inflorescence. L. humile can also disrupt dispersal of seeds by native ants, resulting, for example, in lower survival of seeds after fire in the South African fynbos (Giliomee 1986; Bond & Slingsby 1984). In New Zealand, L. humile will compete strongly with other species for carbohydrate resources, which could have flow on effects for plant pollination. Seed dispersal is unlikely to be affected directly, as seeds are only a small component of the diet of L. humile, and plants with seed adapted for ant dispersal are absent from New Zealand.

12 HarrisPotential impact of the Argentine ant in New Zealand L. humile feed extensively on the honeydew produced by homopterans. They actively disperse the homopterans and protect them from predation to maintain the food source. In New Zealand this may increase adventive homopterans in native habitats, interfere with native predators of homopterans, and aid transmission of diseases between plants. The South Island beech forests, with the abundant honeydew producing scale insect are likely too cold for Argentine ant establishment (see section 6.3). Direct impacts on vertebrates are also possible. Newell and Barber (1913 in Wetterer 1998) described L. humile attacking and killing nesting birds: workers swarm over young chicks in such numbers as to cause their death. Such attacks have been observed in New Zealand (V. Van Dyk pers. comm.). Competition for food is also highly likely and may cause the decline of some species (e.g. Suarez et al. 2000).

4.2 HUMAN AND ECONOMIC IMPACTS

L. humile ranks highly as a domestic nuisance species. They invade houses and are capable of penetrating food containers (Davis & Van Schagen 1993). They infest gardens, making outdoor dining difficult. When nests are disturbed, foragers will run up legs and arms, and some people are sensitive to their bite. L. humile has the potential to carry and hence spread disease (e.g. Staphylococcus, Candida, and Enterococcus) around buildings, including hospitals (Fowler et al. 1993). L. humile has been documented causing economic losses through: The dispersal of homopterans (e.g. scales, aphids) and their protection, which reduced homopteran losses from predation. This reduces the quality of crops and disrupts biological control (e.g. citrusDavis & Van Schagen 1993; coffeeReimer et al. 1993; vineyardsVega & Rust 2001) Holes chewed in plastic drip irrigation pipes that have caused losses in orchards (Chang & Ota 1990) Contamination of food products (Van Schagen et al. 1993) The robbing of bee hives and predation of bees, which affect honey production and pollination industries (Davis & Van Schagen 1993; Vega & Rust 2001) Disruption of the poultry industry through stress on chickens and killing of hatchlings (Davis & Van Schagen 1993) Trade restrictions as a result of contamination of exports to countries without L. humile e.g. China, and Korea (Davis & Van Schagen 1993) Their potential role in the transmission of pathogens from one plant to another through their feeding on sugary exudate (El-Hamalawi & Menge 1996) and the transfer of sap-feeding Homoptera

Science for Conservation 196 13 5. Control options for L. humile

5.1 CHEMICAL CONTROL

With a world-wide distribution and wide-ranging impacts, much effort has been focussed on control strategies. For areas of high conservation significance, such as offshore islands, eradication of populations is more desirable than management requiring ongoing chemical application. For areas of widespread infestation, maintaining low levels of L. humile through ongoing treatment is likely to be the only option. Designing a chemical strategy to suit the behaviour of L. humile is the biggest challenge to successful chemical control. At any one time, only a small percentage (~2%) of ants are foraging (Davis & Van Schagen 1993). The particle size of the toxin is also critical as only very small particles are imbibed (J. Van Schagen pers. comm.). Direct sprays have little effect unless every colony is exposed and treated. If workers feel affected by poison they refuse to feed nest mates (Davis & Van Schagen 1993). Many insecticides are repellent, resulting in unaffected ants staying inside the nest until insecticide residues fall to low levels. With highly persistent insecticides, foragers are eventually forced to forage on contaminated foods. Successful eradication strategies using the persistent sprays dieldrin, chlordane and, later, heptachlor were developed, but these have been withdrawn due to environmental concerns about the chemicals used (Davis et al. 1993). Just before the cessation of heptachlor use in Western Australia, L. humile had been successfully eradicated from 31,000 ha of land, and only 1500 ha remained untreated (including wetland habitat deemed inappropriate for heptachlor treatment). With the cessation of treatment, the area of infestation is again increasing, having doubled to about 3000 ha in less than 3 years (Van Schagen et al. 1993). With the cessation of the use of persistent insecticides, toxic baiting is now considered the most effective control method (Davis et al. 1993). For baits to be effective against ants they must be non-repellent, have a delayed action to allow spread throughout the colony, and be effective over a large range of concentrations (to counter dilution through food exchange) (Stringer et al. 1964). Liquid sucrose has been shown to be the most attractive non-toxic bait to L. humile (Baker et al. 1985). Sucrose solutions containing boric acid are effective at controlling workers in both the laboratory and the field, provided the boric acid concentration is no greater than 1% (Hooper-Bui & Rust 2000; Klotz et al. 2000a). Effective control of queens was achieved in the laboratory when the bait was continuously available, but not when available only for a period of 24 hrs. The main efforts to achieve highly effective baits for L. humile have been the development of protein baits specifically to target queens that are fed protein for egg development (Davis et al. 1993). Two of the most promising toxins, Mirex and sulfluramid, have also been withdrawn from the market, the latter as

14 HarrisPotential impact of the Argentine ant in New Zealand recently as late 2000. Many potential replacements have been tested, and only three shown promise: avermectin, hydramethylnon (Davis et al. 1993), and fipronil. Avermectin was shown to be effective in laboratory trials (Baker et al. 1985), but a successful field formulation has not been produced. Hydramethylnon has been intensively trialled in Western Australia (P. Davis unpublished data) and elsewhere (Knight & Rust 1991; Blachly & Forschler 1996). It is a slow acting stomach poison that takes several days to kill. It is not as effective as Mirex and was not considered to be able to achieve eradication based on laboratory trials (Knight & Rust 1991). Maxforce Granular Insect Bait is a commercially available formulation made from ground silkworm and containing 1% hydramethylnon. It is highly attractive to foragers and reduces worker populations in the field, but failed to achieve eradication of populations in field trials after one or two applications at rates of 4.5 kg/ha (Klotz et al. 1998; Krushelnycky & Reimer 1998a, 1998b). Bait moulding, quick forager mortality, and the rapid UV breakdown of hydramethylnon were cited as reasons for the short exposure time of ants to the bait and the subsequent failure to achieve eradication (Krushelnycky & Reimer 1998b). In the laboratory, significant queen mortality was not achieved compared with an experimental protein bait containing either hydramethylnon or sulfluramid developed by Agriculture Western Australiahereafter called WA Bait (P. Davis unpublished data). Field trials of the WA bait containing sulfluramid achieved eradication after one blanket treatment (small amounts of bait are placed in ant pathways every few metres over the infected area) with a follow-up treatment of any remaining populations after 12 months (P. Davis pers. comm.). Hydramethylnon in the WA bait had not resulted in eradication after two treatments when the trial was stopped. Amdro is another commercially available bait in the US made from soybean oil on corn grit and containing hydramethylnon. It can also reduce L. humile worker abundance (Klotz et al. 2000b), but does not appear to be sufficiently attractive to result in significant queen mortality. Fipronil is a relatively new phenylpyrazole class of neurotoxic insecticides that blocks neurological inhibition. It is fast-acting compared with sulfluramid, but this speed of action does not result in lower effectiveness for Vespula wasp control (Harris & Etheridge 2001). Laboratory trials of fipronil against L. humile compared favourably with other products, including hydramethylnon, killing colonies when used at concentrations as low as 1 ´ 105% (Hooper-Bui & Rust 2000; Costa & Rust 1999). The addition of fipronil (at concentrations of 0.001% and 0.0001%) to sucrose containing a dye marker reduced both the total consumption and the distance the sucrose was spread compared with a non- treatment solution, and this reduction was greater at the higher fipronil concentration (Ripa et al. 1999). The higher the concentration of fipronil used, the less spread there will be between nest mates and colonies. A formulation of Maxforce containing 0.01% fipronil in a pre-packaged bait station is commercially available in the US. Argentine ants are listed on the product label, but I have not found any trial data for this formulation demonstrating its efficacy against this species.

Science for Conservation 196 15 The WA bait containing 0.01% fipronil was trialled in the field just before the removal of sulfluramid from the market. L. humile populations were controlled within the treated area (several hectares), but there was rapid re-invasion from the surrounding areas so an assessment of queen survival was not possible (P. Davis pers. comm.). Baiting strategies using fipronil at concentrations similar to or lower than those trialled in Western Australia are also being developed for fireants (S. invicta Collins & Callcott 1998) and crazy ants (Anoplolepis gracilipesD. Slip unpublished data.). Control of ant populations in plant nurseries is an important step to reduce the rate of L. humile spread. Soil mix treatments of fipronil at 5 ppm active ingredient (AI) prevented ants from establishing in pots and killed all workers and queens when colonies were forced to inhabit them (Costa & Rust 1999). Fipronil broadcast onto the soil surface at rates of 14 g (AI)/ha killed all queens within the pot but took 8 weeks to achieve 100% mortality. A new area of research is investigating the use of trail pheromones to increase bait attraction (Greenberg & Klotz 2000). Addition of L. humile trail pheromone to a sucrose solution increased consumption of the liquid in both the laboratory and the field. A commercial product using this technology is some way off. If baiting is not an option, L. humile can be chemically excluded from specific sites (e.g. tree trunks): the use of cotton twine permeated with 40 mg of farnesol and stickem per cm of twine will exclude foragers for 23 months (Shorey et al. 1996).

5.2 BIOLOGICAL CONTROL

Phorid flies, Pseudacteon spp. (Diptera: Phoridae), attack L. humile in Brazil and deter ant foraging during the flies diurnal activity period (Orr & Seike 1998). Biological control of L. humile has not been attempted within its introduced range. In the US, phorid flies of the same genus (Pseudacteon spp.) are being considered for release against the red imported fire ant S. invicta (Porter & Alonso 1999). Pathogens have also been trialled against S. invicta (Stimac et al. 1990), but the ants, like other social insects, have very efficient hygienic behaviour that inhibits the spread of pathogens that appear in the nest. This behaviour has so far prevented the use of pathogens for biological control of many social insects. An alternative approach to biological control would be to increase the genetic diversity of L. humile through the importation of queens or males. The aim would be to reproduce the situation in their native range where intra-specific differences are recognised and there is as a consequence, wider territorial spacing of colonies and significantly lower ant densities (Tsutsui et al. 2000). A novel control technique being used by a student at the University of California, Davis, to control L. humile was baiting with a hay-bale (B. Inouye pers. comm.). The ants moved their nests out of wetter or less sheltered areas and into the hay, which was then burnt or frozen.

16 HarrisPotential impact of the Argentine ant in New Zealand 6. L. humile in New Zealand

6.1 NEW ZEALANDS ANT FAUNA

New Zealand, like other oceanic islands, has a depauperate ant fauna (Berry et al. 1997; Reimer 1993), with only 10 endemic species and possibly two other natives (Valentine & Walker 1991; Don unpublished in Berry et al. 1997; J. Berry pers. comm.). An additional 2224 adventive species have also established (Harris & Berry 2001; Valentine & Walker 1991). The endemic species are generally widespread geographically, mainly in forested habitats. An exception is Heteroponera brouni Forel, which appears to be restricted to northern North Island forests (Brown 1958). It is clear from the scarcity of literature on ants in New Zealand that details on the distribution of most adventive species are incomplete. For several species there are no recent collection records, and it is unclear whether they have successfully established. The occurrence of adventive species in native systems is poorly documented. Other adventive ants established in New Zealand are unlikely to restrict the distribution of L. humile, with the possible exception of P. megacephala. However, P. megacephala is currently restricted to a few suburbs around Auckland (Berry et al. 1997), and as it requires much warmer temperatures than L. humile (Reimer 1993), is unlikely to become widespread. The red imported fire ant (S. invicta), a nest of which was discovered near Auckland airport in March 2001, is another species that could restrict L. humile distribution. However, if it became established here, it would undoubtedly surpass L. humile as New Zealands worst ant pest, with consequences similar to those of L. humile for native ecosystems (Porter & Savignano 1990), as well as major human health and agricultural impacts (Barr & Drees 1996; Oi et al. 1994).

6.2 L. humile ARRIVAL AND SPREAD

L. humile was first recorded in Auckland in 1990 (Green 1990), and was already establishing across several hectares. There was no attempt to eradicate the species at that time, as it was already well established (Green 1990). Over the following 6 years there were few L. humile collected and no formal reports of its spread. In summer 1997/98, two discrete populations of L. humile in Mt Maunganui were surveyed (Osborne 1998). Recent publicity, and increased searching for the species as part of this project, has produced many more records. Currently, L. humile is known from numerous sites in northern North Island and two cities in the South Island (Fig. 2). However, as L. humile is plain brown, relatively small, and displacing other pest ant species in many urban areas, people often are not aware of its presence until large numbers build up. Therefore, the full extent of its current distribution will be underestimated. A nation-wide survey of L. humile, predominantly in modified systems, has been funded by MAF Biosecurity, and is due to be completed by 30 June 2001. Data

Science for Conservation 196 17 Figure 2. Sites in the North and South Islands of New Zealand where L. humile has been confirmed up to 28 March 2002.

collected as part of this project have been made available for addition to that survey. The spread of L. humile into native systems is unlikely to be detected unless ants are specifically being monitored, as is currently being instigated on offshore islands in the Northland Conservancy, DOC (A. Booth pers. comm.). L. humile was recently discovered to be widespread in Whangarei (pers. obs.). Raising awareness of L. humile among DOC Northland staff resulted in the discovery of L. humile at the Whangarei field centre (sample sent to me by A. Booth). This is a significant find as field gear for offshore islands is stored there. The field centre has been treated and a policy put in place to reduce chances of transporting ants to offshore islands through regular monitoring and retreatment as necessary (A. Booth pers. comm.). At Mt Maunganui, the initial survey of L. humile during summer 1997/98 (Osborne 1998) was repeated in summer 2000/01 (Dykzeul 2001). During this period, the number of separate infestations (unconnected and representing jump-dispersal events) increased from two to six, and the total area infested from 21.0 to 43.8 ha. Surprisingly, for no obvious reason, one of the original infestations had decreased in size (down from 11.4 ha to 5.2 ha).

18 HarrisPotential impact of the Argentine ant in New Zealand 6.3 POTENTIAL FUTURE DISTRIBUTION IN NEW ZEALAND

6.3.1 Methods

Northland falls into the latitude belt considered ideal for L. humile (3036o latitude). However, L. humile is well established in and around Auckland (about 37o latitude), in urban Wellington (about 41.3o latitude), and there is a small population in urban Christchurch (about 43.5o latitude). Many ant species can survive in urban areas below their lower temperature limits due to the warmer microhabitats created by urban settlements (e.g. L. humileSuaraz et al. 2001, Monomorium pharonis (L.)Edwards 1986, Paratrechina longicornis (Latreille)Dubois & LaBerge 1988).

New Zealands mean annual temperatures range from 5.12oC to 16.27oC (Leathwick & Stephens 1998). Lower temperature limits were estimated using temperature data from the colder New Zealand sites where L. humile has established (Table 2) and taking into account data on temperature limitations (section 3.5.1). No sites in New Zealand are considered too hot for L. humile as they are established in much warmer localities elsewhere (e.g. California). As a check of the assumed temperature limits, temperature data for the more temperate cities in Australia that have infestations of L. humile (see Shattuck 1999) were compared with New Zealand cities (CLIMEXSkarratt et al. 1995).

TABLE 2. MEAN ANNUAL TEMPERATURE IN THE COLDEST AREAS WHERE L. humile HAS BEEN FOUND TO DATE.

HABITAT LOCATION MEAN ANNUAL TEMPERATURE (oC)

Urban Riccarton, Christchurch 11.9 Kelburn, Wellington 11.9 Port Nelson 12.8 Lower Hutt 12.8 Petone 12.6

Scrub Tiritiri Matangi Island 15.2 Te Whau Point, Blockhouse Bay 14.8 Piha 13.8

Australiaurban Launceston 12.2 Hobart 12.6

L. humile has been present in Christchurch in a localised area for several seasons but dispersal appears slow. The coldest sites where L. humile has been reported in Australia have higher mean annual temperatures than the Riccarton and Kelburn infestations (Table 2). There are currently few New Zealand sites away from urban development where L. humile have been found, and these are all in northern sites (Table 2). Using these data, I have divided areas into temperature bands: T2) above 12oC ideal for establishment in urban areas; T1)

Science for Conservation 196 19 10.5oC to 12oC; and T0) below 10.5oC, assumed to be too cold. I also assume that L. humile can persist in urban areas 1.5oC below the temperature limits in other habitat types (Table 3). Based on the information on habitat associations (section 3.5.2), I have used the New Zealand Land Cover Database B1 classes and divided them into 3 habitat groups (H2 = suitable, H1 = marginal, and H0 = unsuitable). Where no data are available to indicate habitat suitability, the habitat was given a H1 classification (marginal). A geographic information system (GIS) was used to overlay mean annual temperature data (Leathwick & Stephens 1998) and land-cover data using the assumptions in Table 3 to obtain an estimate of the potential future distribution of L. humile. New Zealand is divided into five risk-establishment categories should L. humile reach a location. 1 T2+H2 High Risk 2 T2+H1 Moderate Risk 3 T1+H2, T1+H1 Low Risk 4 H0+T2, H0+T1 Unsuitable habitat 5 T0+H2, T0+H1, T0+H0 Too cold

TABLE 3. ASSUMED HABITAT PREFERENCES AND TEMPERATURE LIMITATIONS (MEAN ANNUAL TEMPERATURE) USED TO ESTIMATE THE POTENTIAL FUTURE DISTRIBUTION OF L. humile.

HABITAT HABITAT TEMPERATURE BANDS FOR GROUP HABITAT GROUP (oC) NZLCDB1 LAND COVER CLASSES T2 T1 T0

1 Urban Area H2 >12 10.512 <10.5 2 Urban open space H2 >12 10.5-12 <10.5 3 Mines and dumps H1 >13.5 1213.5 <12 4 Primarily Pastoral H1 >13.5 1213.5 <12 5 Primarily Horticultural H2 >13.5 1213.5 <12 6 Planted forest H1 >13.5 1213.5 <12 7 Riparian Planting H1 >13.5 1213.5 <12 8 Major Shelterbelts H1 >13.5 1213.5 <12 9 Tussock grassland H1 >13.5 1213.5 <12 10 Scrubland H2 >13.5 1213.5 <12 11 Mangrove H1 >13.5 1213.5 <12 12 Indigenous forest H0 >13.5 1213.5 <12 13 Bare Ground H1 >13.5 1213.5 <12 14 Inland Wetland H1 >13.5 1213.5 <12 15 Coastal wetland H1 >13.5 1213.5 <12 16 Coastal sand H0 >13.5 1213.5 <12 17 Inland Water H0 >13.5 1213.5 <12 18 Unclassified H1 >13.5 1213.5 <12

20 HarrisPotential impact of the Argentine ant in New Zealand 6.3.2 Results Under the assumed criteria, most of the South Island and inland North Island will be too cold for L. humile to establish outside urban areas (Figs 3A and B, see next pages). However, large areas of northern and coastal North Island are considered high risk for L. humile establishment. Many offshore islands are high risk, including islands such as Little Barrier Island with high valued faunas. Geothermal areas in the centre of the North Island may support L. humile where surrounding habitat is too cold. The prediction can be fine-tuned as more information becomes available on local invasion by L. humile. Over the next 10 years it is likely that infestations will be detected in many more urban areas, allowing temperature limitations for this habitat to be refined. Dispersal into native habitats will take considerably longer. Localities with low visitation rates, especially by boat or vehicle, may never have colonies transported into the area, and with a natural dispersal rate of around 150 m/yr, (Sueraz et al. 2001) could remain free of L. humile for many centuries. For example, it would take over 600 years for L. humile to spread from Kaitaia to Cape Reinga via budding. So, although the risk of L. humile establishing at many sites in northern New Zealand is high, the risk of their dispersal to them is low. In the future, the distance from urban areas and sites with L. humile could be added as additional data layers to Fig. 3. This would then combine risk of establishment and risk of dispersal to identify those sites most likely to be invaded next. Forest habitat has been assumed not to be at risk of invasion. The best place to confirm this assumption in the future will be areas of forest in northern New Zealand that are close to urban areas with well-established L. humile populations. Such sites include the Waitakere Ranges and forested areas on the outskirts of Whangarei. Fragmentation of forest will likely open up the edges of these habitats to invasion, as the ants may be able to forage into the edge during the height of summer, even if the habitat does not support colonies year round. Also climatic condition on the edge of a forest fragment will be intermediate between the forest interior and open habitat (Davis-Colley et al. 2000).

Science for Conservation 196 21 Figure 3A. Prediction of areas of the North Island which are suitable for L. humile invasion. The location precision is appropriate for publication at 1:50 000 and the minimum mapping unit is one hectare. A much larger (700 ´ 900 mm) version of this map is available for viewing from the author.

22 HarrisPotential impact of the Argentine ant in New Zealand Figure 3B. Prediction of areas of the South Island which are suitable for L. humile invasion. The location precision is appropriate for publication at 1:50 000 and the minimum mapping unit is one hectare. A much larger (700 ´ 900 mm) version of this map is available for viewing from the author.

Science for Conservation 196 23 7. Tiritiri Matangi Island eradication trial

This trial is being conducted in collaboration with Dr Chris Green, Auckland Conservancy, Department of Conservation. Successful eradication will not only have major benefits for Tiritiri Matangi Island, but will provide a strategy for use on other offshore islands or key mainland sites should L. humile become established.

7.1 BACKGROUND

L. humile was first identified on Tiritiri Matangi on 29 March 2000. The infestation was already well established around the main jetty and may have originated from machinery brought to the island during the construction of the new jetty in 1998. A second small infestation at Northeast Bay is known to have originated in December 1999 when a dinghy was moved from within the main infestation to Northeast Bay (~1.5 km to the north of the main infestation). In January 2001, just before poisoning, the boundaries of the infestation were determined using a mixture of non-toxic baiting and hand searching. The main infestation, which is centred around the jetty, was 9.3 ha, and the small infestation at Northeast Bay was 0.5 ha (both areas include a buffer zone of about 20 m around the outside where no L. humile foragers were found) (Fig. 4). These sizes are two- dimensional and do not account for the terrain. Accounting for the steep terrain in parts of both infestations increases the total treated area to about 11 ha. The WA bait, a protein-based matrix, was considered the best available option to use. Landcare Research negotiated an agreement with Agriculture WA to allow the bait to be trialled in New Zealand. The bait formulation is not specific to L. humile, but

Figure 4. Location of L. humile infestations on Tiritiri Matangi Island.

24 HarrisPotential impact of the Argentine ant in New Zealand should they be present, few, if any, other ant species occur, and L. humile foragers will monopolise the bait, thus reducing the impact on other invertebrates. The first treatment was originally planned for spring 2000. However, just before this, the insecticide sulfluramid, the toxin we originally planned to use in the bait, was withdrawn from the market. Fipronil at 0.01% was substituted as an alternative as it was being adopted widely in ant control strategies and available evidence suggested use of hydramethylnon would not result in eradication. An application was made to MAF for provisional registration of the bait matrix containing fipronil and this was obtained on 24 January 2001.

7.2 METHODS

7.2.1 Poison baiting The outer edge of the infestation (including the buffer) was marked with flagging tape. Teams of 47 people moved in lines 3 m apart and placed bait every 2 m. Bait dries rapidly and becomes unpalatable if exposed to full sun, so portions of bait (about 2 g) were placed on the ground in shady positions, except for highly exposed coastline and open grassed areas which were treated after dark. Baiting began on 30 January at the northern edge of the main infestation. The outer edge of each teams line was marked as a guide for the next team. Light misty rain fell on 31 January so no bait was put out to avoid any bait being washed away. On 1 February the Northeast Bay infestation was treated first, then baiting continued within the main infestation until all bait was used. The area north of the dotted line on Fig. 5 marks where baiting was completed at that time. On 14 February the remaining area (~3.05 ha) was treated. The shingle on the edge of the roadside, an area of high ant density, was treated after dark. Treating the grassed areas after dark also avoided Pãkeko (Porphyria porhyria) that foraged in these areas. The only endangered bird species occasionally found in the area being treated, a pair of Takahe (P. manteli), were penned during the baiting operation.

Figure 5. Location of non-toxic bait monitoring lines within (n=7) and outside (n=2) the main L. humile infestation on Tiritiri Matangi Island. The edge of the infestation is marked by the solid black line. The area north of the dotted line was the area treated during the first baiting period (30 January and 1 February), the remainder was treated on 14 February 2001.

Science for Conservation 196 25 7.2.2 Ant monitoring To monitor changes in ant numbers (L. humile and other species) over time, a series of monitoring lines, each with 5 or 10 bait stations, were established inside and outside the infestations (see Fig. 5 for monitoring lines around main infestation, a monitoring line was also established at Northeast Bay, not shown on Fig. 5). Lines were established non-randomly along tracts so that monitoring could be done during a one day visit to the Island. Each bait station consists of a vial containing ~5 g of non-toxic protein bait (the base of the toxic bait) placed on the ground. Baits were left out for 3 hours and the lid then placed on the vial to trap the ants on the bait. Monitoring lines were run at least once before poisoning and 24 times after poisoning from February to April 2001. James Haw, a PhD student at Auckland University, is measuring the effect of ant removal on the invertebrate fauna and carrying out two independent measures of ant populations. He is measuring ant trail activity on randomly selected tree trunks at two sites within the treatment area and setting non-baited pitfall traps. Although pitfall traps are not very effective at catching L. humile, they may catch some of the other ant species as they re-invade the area previously dominated by L. humile.

7.3 PRELIMINARY RESULTS AND DISCUSSION

7.3.1 Bait application It took about 179 person hours to treat the infested area (~18 hrs/ha). A total of 61.9 kg of bait was applied (6.3 kg/ha when correcting for the terrain). The application rate was about double the planned rate, partially due to the undulating terrain (the application rate was 5.8 kg/ha assuming 11 ha was treated when correcting for the undulating terrain). The bait was highly attractive to L. humile, and a large numbers of foragers were seen feeding within a few minutes of it being placed near an active trail. Activity on baits ceased within 12 hours of baiting. Bait application times and bait usage were high compared with baiting in an urban area where the terrain is generally flat and large areas of the infestation are not treated (e.g. concrete slabs or buildingsunpublished data). At Port Nelson, a similar sized infestation was treated and took 4.5 hrs/ha and used 2.5 kg of bait/ha.

7.3.2 Bait effectiveness Immediately before poisoning, an average of 295 ± 57 (mean ± SE) L. humile per bait station was recorded on baits within the infested areas. Fewer L. humile were sampled at Northeast Bay compared with the main infestation (Table 4). L. humile numbers at these stations dropped to an average of 0.06 ± 0.04 ants per station 12 to 15 days after poisoninga 99.98% reduction in numbers. It is assumed most of the reduction in numbers was due to baiting, as reductions in numbers due to the onset of winter will occur slowly and later in the season than the poisoning. Ten days after the initial poisoning, it was evident that the toxin was still having some effect, as dying ants and abandoned dead brood

26 HarrisPotential impact of the Argentine ant in New Zealand TABLE 4. EFFECT OF TOXIC BAITING ON L. humile NUMBERS.*

L. humile CATCH (MEAN ± SE)

BAIT PRE-POISON PRE-POISON POST-POISON POST-POISON POST-POISON POST-POISON LINE 4750 DAYS 14 DAYS 03 DAYS 1215 DAYS 3652 DAYS 6480 DAYS

BA 131. 6 ± 35.8 413.0 ± 35.3 0 ± 0 0.2 ± 0.2 0 ± 0 CT 500.0 ± 53.1 0.2 ± 0.2 0 ± 0 0 ± 0 0 ± 0 FA 91.8 ± 56.6 215.4 ± 36.9 0.2 ± 0.2 0 ± 0 0 ± 0 GA 174.3 ± 50.1 401.3 ± 51.5 0 ± 0 0.2 ± 0.2 0 ± 0 RA 279.3 ± 64.1 0 ± 0 0 ± 0 WA 135.9 ± 43.0 172.8 ± 45.4 26.4 ± 12.6 0 ± 0 0 ± 0 0 ± 0 NE 81.2 ± 42.4 0 ± 0 0 ± 0 0 ± 0

* Non-toxic protein baits were placed out once for 3 hours per bait line on each monitoring date. See Fig. 5 for location of monitoring lines, except NE, which is at Northeast Bay.

were observed. After poisoning, very few foraging trails could be seen around the treated area. On 22 March, both infestations were inspected in detail. A few aggregations of workers were found within the main infestation and one active colony had apparently missed treatment. The active colony was near the mean high water line in an area of high ant densities and there may have been a relatively low amount of bait in the vicinity. At Northeast Bay, only one small colony was found. After a worker was first seen it took about 30 minutes searching over an area of about 2 square metres to find the colony. The colony was sprayed with Ant-Ban® and there was no activity in the area of the colony when it was rechecked on 19 April. On 19 April, no L. humile were sampled on the monitoring baits, but there was visual evidence of increased forager activity in some areas of the main infestation. No foragers were recorded on a series of monitoring tree trunks at one site within the main infestation, near monitoring line CT, while at a second site within the main infestation, between monitoring lines CT and WA, the numbers had begun to increase (James Haw pers. comm.).

7.3.3 Evaluation of non-toxic baits for monitoring Before poisoning, the non-toxic protein baits within the main infestation area had attracted high numbers of L. humile (Table 4), and 89% of baits attracted foragers after 3 hours. At Northeast Bay, fewer bait stations attracted L. humile (40%), reflecting their patchy distribution and relatively low abundance at this site. The monitoring baits have adequately reflected the dramatic reduction in L. humile numbers seen by visual inspection of nest sites and trunk activity. To confirm the absence of L. humile from areas, after further treatment, more monitoring baits would be needed and these could be left out for longer periods and checked at intervals for sign of L. humile. Intensive hand searching would also be required to confirm eradication, as we do not currently have a more effective monitoring method for picking up ants at very low densities. Copper skinks (Oligosoma aenea) were seen taking non-toxic bait and had removed all bait from several bait stations during monitoring in March and

Science for Conservation 196 27 April. This was not a problem encountered earlier, and may reflect the time of year and/or the removal of L. humile, freeing up access to the baits. Mesh entrance covers will be trialled to exclude the skinks from bait stations. Monitoring using non-toxic protein baits will be most effective during spring and summer. In autumn, the protein bait matrix may have reduced attractiveness to L. humile as there will be a lower demand for protein within the colony. At a non-treatment site in Mt Maunganui, non-toxic bait stations sampled fewer L. humile foragers in April despite there being no visible decrease in forager activity (unpublished data). The use of jam baits alongside protein baits would indicate if either bait is more attractive at this time of year or if colonies are in decline and less attracted to bait in general as the temperature drops.

7.3.4 Other ants Only 5 of the 11 other ant species recorded on Tiritiri Matangi have been sampled on protein monitoring baits (Table 5). These are in low numbers compared to L. humile, even outside the infestation. Other species sampled in the treated area were on monitoring lines at the edge of infestation, where L. humile has not yet monopolised the area and excluded other species (BA, top end of WA, Northeast Bay). Since treatment, the numbers of Mayriella abstinens on baits within the margin of the main infestation have increased. This species was recorded in the area before poisoning (unpublished data) but was not sampled on protein baits, probably due to exclusion by L. humile. Treatment of Northeast Bay appears to have negatively affected Tetramorium grassi. Neither Monomorium antarcticum nor Pachycondyla castanea were sampled in the areas of L. humile infestation, despite being the most abundant species on the edge of the main infestation. These native species may be useful as indicators of recovery of the invertebrate community within the infested area should eradication of L. humile be successful. Several queens of Iridomyrmex anceps (Roger) and a worker of Ochetellus glaber (Mayr) have been seen in the middle of the main infestation since poisoning, but may not be sampled by the protein baits.

TABLE 5. SUMMARY OF ANTS SAMPLED FROM PROTEIN BAITS ON TIRITIRI MATANGI ISLAND DURING 2000/01.

MAIN OUTSIDE MAIN NORTHEAST INFESTATION INFESTATION BAY

PRE- POST- PRE- POST- PRE- POST- SPECIES TREATMNT TREATMNT TREATMNT TREATMNT TREATMNT TREATMNT

L. humile (Mayr) (a) 245.3 0.03* 0 0 81.2 0* Mayriella abstinens Forel (a) 0 2.04* 0 0 0 0 Monomorium? antipodum 0.09 0.38 0 0 0 0 Forel (a?) M. antarcticum (Smith) (n) 0 0 9.50 1.90 0 0 Pachycondyla castanea 0 0 0.30 0.43 0 0 (Mayr) (n) Tetramorium grassii 0.09 0.04 1.70 1.23 9.10 0.43* Emery (a)

* Difference between pre- and post-poison numbers statistically significant (P < 0.05). n = native, a = adventive

28 HarrisPotential impact of the Argentine ant in New Zealand 7.3.5 Overview of strategy and future plans The results on Tiritiri Matangi are consistent with results from trials in Nelson and Mt Maunganui. L. humile foragers actively feed on toxic bait for the first few hours. This is followed by a dramatic reduction in ant numbers within a short timeframe (<24 hours). Detailed searches reveal that while whole colonies, including queens, are being destroyed across much of the treated area, some colonies survive. This indicates that the bait is highly effective at killing colonies but that not all colonies are getting sufficient bait. Foragers may be dying too quickly and therefore the fipronil is not getting spread to every colony or to every individual within colonies. Achieving 100% kill with one treatment is not a realistic expectation, especially considering the terrain and initial density of ants. However, the Northeast Bay infestation was relatively small and the core infestation relatively easy to treat compared with the main infestation. Despite this at least one queen still survived at Northeast Bay. To succeed in achieving eradication all the surviving queens need to be killed. With the onset of winter, L. humile numbers will decline further. The best time for the next treatment will be next spring, when surviving colonies begin to expand and protein is again in demand. The plan is to treat the whole infestation on Tiritiri Matangi once more and then switch to intensive monitoring and specific treatment of any remaining areas of L. humile activity. Results were highly encouraging compared to attempts to control Argentine ants overseas using Maxforce Granular Insect, which to have limited impact on queens (Klotz et al. 1998; Krushelnycky & Reimer 1998a, 1998b). The results were similar to those achieved with the toxin sulfluramid in the same bait matrix, in Western Australia, with ant numbers being dramatically reduced and only a few remnant populations left in the treated area after one application. The main difference was the speed with which ants were killed with the fipronil. Two alterations could be made to the initial baiting strategy that might increase the likelihood of all ants being killed. First, an increase in the number of baits per unit area and/or second, a decrease in the fipronil concentration of the bait so that it takes longer to kill and there is a greater exchange of the bait between nests. This assumes that the limitation to success is not the presence of some form of physical or behavioral refugia by which some ants are never put at risk by the control method (J. Parkes pers. comm.). If this were the case, an additional and different method would be required The bait is currently being placed on a 2 ´ 3 m grid. Reducing the 3 m spacing between the applicators would greatly increase the time to treat the infestation. However, bait could be placed on a 1 ´ 3 m grid without a significant increase in effort. Care needs to be taken in dense vegetation to keep the 3 m spacing as a maximum to avoid situations where colonies are over 2 m from bait. A reduction in concentration of fipronil in baits would be likely to increase the time workers could forage until affected by the toxin and would allow more interchange of food between nest-mates and between nests, as demonstrated by Ripa et al. (1999). Fipronil displays toxicity over a wide range of concentrations (Hooper-Bui & Rust 2000), so it is unlikely that efficacy would be reduced.

Science for Conservation 196 29 Cessation of foraging trails occurs within 12 hours using 0.01%. When sulfluramid was trialled, the total cessation of foraging activity took considerably longer than with fipronil (weeks compared with hours), although some reduction in forager activity was seen within 48 hours (unpublished data, Rupes et al. 1997). Comparative trials of 0.01% and 0.001% fipronil formulations are needed to determine if efficacy is improved.

8. Conclusions

L. humile is a highly invasive opportunistic species that is now established in many countries. It is highly suited to disturbed and highly modified habitats, and being dispersed through human activities. L. humile is still spreading in New Zealand. Excluding urban areas, most areas of highest risk of establishment are in northern New Zealand. Urban areas are likely be invaded first, but L. humile has the potential to establish in low stature scrub/coastal vegetation such as that on Tiritiri Matangi Island. Many other offshore islands are at risk of invasion. Stands of indigenous forest habitat are unlikely to be invaded. Small fragments may be at risk if colonies live on the forest margin and forage into the fragment, or disperse into the fragment over the summer. Dispersal is slow if unaided by humans (~150 m/yr), and L. humile will take hundreds of years to spread throughout suitable habitat in New Zealand. Areas of high public visitation and landing sites on offshore islands are likely to be the first sites on the conservation estate to be invaded and would be the sites to monitor for signs of L. humile arrival. Wherever L. humile establishes in native vegetation, the total ant biomass at that site will increase and there will be detrimental impacts on the native fauna, particularly the invertebrate community, with many species declining in numbers or becoming locally extinct. It is difficult to quantify the impact of L. humile in native ecosystems in New Zealand at this time, as there are few sites where native vegetation has been invaded. James Haw, a PhD student at Auckland University, is currently investigating the impact of L. humile by comparing invertebrate communities at several sites on the urban fringe with and without L. humile and also monitoring the effect of L. humile removal on the invertebrate fauna on Tiritiri Matangi Island. Although limited by the nature of disturbed urban edge habitats available to him, and the short timeframe for monitoring invertebrate populations before treatment on Tiritiri Matangi Island, the study will provide the first data on impacts of L. humile in New Zealand. There is a range of options available, or under development, for control of L. humile, but most options do not have the ability to eradicate whole populations. It is unlikely that better initial results would have been achieved on Tiritiri Matangi with any other product currently available. However, there are still queens present on Tiritiri Matangi that need to be killed to achieve eradication.

30 HarrisPotential impact of the Argentine ant in New Zealand A reduction in toxin concentration and an increase in bait density may achieve even greater reductions in L. humile densities than those achieved with the first trials on Tiritiri Matangi and two urban populations. Poisoning L. humile over areas greater than several hectares using existing methods is labour intensive and hence expensive. The earlier an infestation is detected the easier it will be to treat.

9. Recommendations

To reduce the impact of L. humile in natural ecosystems, protocols need to be in place to minimise the chances of transporting them to key conservation areas, and an early-warning monitoring scheme established in areas identified as high risk. A detailed survey of ants in forest patches in and around Auckland and Whangarei should be conducted to confirm the assumption that forest habitat is not suitable for L. humile invasion. The trial on Tiritiri Matangi needs to continue in order to determine if eradication of entire populations can be achieved with current methodology. For the second treatment of Tiritiri Matangi, fipronil concentration should be reduced 10-fold and bait density increased (3 ´ 1 m or 2.5 ´ 1 m). Less bait should be used at each site (~1 g) as L. humile densities will be far lower after the first poisoning. Intensive monitoring of is required for at least two years after the last Argentine ant is sampled to confirm eradication and monitor the re-invasion of other ant species. Additional field trials at another site are needed to determine the impact of a ten-fold and one-hundred-fold reduction in the concentration of fipronil on the amount of bait consumed, the distance bait is spread, and the resulting level of control. A direct field comparison of fipronil with other potential toxins (e.g. hydramethylon) is recommended if eradication is not achieved on Tiritiri Matangi by the end of summer 2003. However, current evidence suggests none of the available alternative toxins offer greater potential (see section 5.1). With the implementation of the HSNO Act, considerable costs will be involved in obtaining provisional registration for field trials and subsequent full registration.

10. Acknowledgements

My thanks to: M. Anne Sutherland who did the GIS manipulations to produce the maps of invasion risk; Chris Green for initiating this project and all his work to get the trial on Tiritiri Matangi underway; all those who have sent me ants to

Science for Conservation 196 31 check (especially Viv Van Dyk); Jo Rees for all the ant counts and bait making; everyone who helped out with the bait application on Tiritiri Matangi; Rachel Standish, Richard Toft, Jacqueline Beggs, and Anne Austin for comments on the draft report. FRST funded the involvement of Landcare Research staff in the eradication trial through the Invasive Invertebrates in Natural Ecosystems programme (Contract C09X011). (Text originated as Landcare Research Contract Report: LC0001/103, DOC investigation no. 3291.)

11. References

Baker, T.C.; Vorhis-Key, S.E. van; Gaston, L.K. 1985: Bait-preference tests for the Argentine ant (Hymenoptera: Formicidae). Journal of Economic Entomology 78: 10831088.

Barr, C.L.; Drees, B.M. 1996: Texas cattle producers survey: impact of red imported fire ants on the Texas cattle industry. Texas Agricultural Extension Service, Texas A&M University, College Station, TX. 226 p.

Berry, J.A.; Green, O.R.; Shattuck, S.O. 1997: Species of Pheidole Westwood (Hymenoptera, Formicidae) established in New Zealand. New Zealand Journal of Zoology 24: 2533.

Blachly, J.S.; Forschler, B.T. 1996: Suppression of late-season Argentine ant (Hymenoptera, Formicidae) field populations using a perimeter treatment with containerised baits. Journal of Economic Entomology 89: 14971500.

Bolger, D.T.; Suarez, A.V.; Crooks, K.R.; Morrison, S.A.; Case, T.J. 2000: Arthropods in urban habitat fragments in southern California: Area, age, and edge effects. Ecological Applications 10: 12301248.

Bond, W.J.; Slingsby, P. 1984: Collapse of an ant-plant mutualism: the Argentine ant (Iridomyrex humilis) and myrmecochorous Proteaceae. Ecology 65: 10311037.

Brown, W.L. jr 1958: A review of the ants of New Zealand. Acta Hymenopterologica 1: 150.

Buys, B. 1987: Competition for nectar between Argentine ants (Iridomyrmex humilis) and honeybees (Apis mellifera) on black ironbark (Eucalyptus sideroxylon). South African Journal of Zoology 22: 173174.

Cammell, M.E.; Way, M.J.; Paiva, M.R. 1996: Diversity and structure of ant communities associated with oak, pine, eucalyptus and arable habitats in Portugal. Insectes Sociaux 43: 3746.

Chang, V.; Ota, A.K. 1990: Ant control in Hawaiian drip irrigation systems. Pp. 708715 in Vander Meer, R.K.; Jaffe, K.; Cedeno A. (Eds) Applied myrmecologya world perspective. Westview Press, Boulder, CO.

Chen, J.S.C.; Nonacs, P. 2000: Nestmate recognition and intraspecific aggression based on environmental cues in Argentine ants (Hymenoptera: Formicidae). Annals of the Entomological Society of America 93: 13331337.

Cole, F.R.; Medeiros, A.C.; Loope, L.L.; Zuehlke, W.W. 1992: Effects of the Argentine ant on arthropod fauna of Hawaiian high-elevation shrubland. Ecology 73: 13131322.

Collins, H.L.; Callcott, A.M.A. 1998: Fipronilan ultra-low-dose bait toxicant for control of red imported fire ants (Hymenoptera, Formicidae). Florida Entomologist. 81: 407415.

Costa H.S.; Rust M.K. 1999: Mortality and foraging rates of Argentine ant (Hymenoptera: Formicidae) colonies exposed to potted plants treated with fipronil. Journal of Agricultural & Urban Entomology 16: 3748.

Davis, P.; Van Schagen, J. 1993: Effective control of pest ants. Journal of Agriculture, Western Australia 34: 9295.

32 HarrisPotential impact of the Argentine ant in New Zealand Davis, P.R.; Van Schagen, J.J.; Widmer, M.A.; Craven, T.J. 1993: A review of Argentine ant research in Western Australia. 195. Perth, WA, Agricultural Protection Board of Western Australia.

Davies-Colley, R.J.; Payne, G.W.; van Elswijk, 2000: Microclimate gradients across a forest edge. New Zealand Journal of Ecology 24: 111121.

De Kock, A.E. 1990: Interactions between the introduced Argentine ant, Iridomyrmex humilis Mayr, and two indigenous fynbos ant species. Journal of the Entomological Society of Southern Africa 53: 107108.

Dubois, M.B.; LaBerge, W.E. 1988: Annotated list of ants in Illinois (Hymenoptera: Formicidae). Pp. 133156 in Trager, J.C. (Ed) Advances in myrmecology. E. J. Brill, Leiden.

Dykzeul, P. 2001: Argentine ant (Linepithema humile): delimiting survey of the Mount Maunganui area. Unpublished Environment BOP operations report 20001/03.

Edwards, J.P.1986: The biology, economic importance, and control of the pharaohs ant Monomorium pharaonis (L.). Pp. 257271 in Vinson, S.B. (Ed.) Economic impact and control of social insects. Praeger Publishers, New York.

El Hamalawi, Z.A.; Menge, J.A. 1996: The role of snails and ants in transmitting the avocado stem canker pathogen, Phytophthora citricola. Journal of the American Society for Horticultural Science 121: 973977.

Folgarait, P.J. 1998: Ant biodiversity and its relationship to ecosystem functioning: a review. Biodiversity & Conservation 7: 12211244.

Fowler, H.G.; Bueno, O.C.; Sadatsune, T.; Montelli, A.C. 1993: Ants as potential vectors of pathogens in hospitals in the state of Sao Paulo, Brazil. Insect Science and its Application 14: 367370.

Giliomee, J.H. 1986: Seed dispersal by ants in the Cape flora threatened by Iridomyrmex humilis (Hymenoptera: Formicidae). Entomologia Generalis 11: 217219.

Green, O.R. 1990: Entomologist sets new record at Mt Smart for Iridomyrmex humilis established in New Zealand. Weta 13: 1416.

Greenberg, L.; Klotz, J.H. 2000: Argentine ant (Hymenoptera : Formicidae) trail pheromone enhances consumption of liquid sucrose solution. Journal of Economic Entomology 93: 119122.

Harris, R.J.; Berry, J.A. 2001: Confirmation of the establishment of three adventive ants (Hymenoptera: Formicidae) in New Zealand: Cardiocondyla minutior Forel, Ponera leae Forel, Mayriella abstinens Forel. New Zealand Entomologist 24: 5356.

Harris, R.J.; Etheridge, N. 2001: Comparison of baits containing fipronil and sulfluramid for the control of wasps (Vespula spp.). New Zealand Journal of Zoology 28: 3948.

Haskins, C.P.; Haskins, E.F. 1988: Final observations on Pheidole megacephala and Iridomyrmex humilis in Bermuda. Psyche Cambridge, Mass 95: 177184.

Hee, J.J.; Holway, D.A.; Suarez, A.V.; Case, T.J. 2000: Role of propagule size in the success of incipient colonies of the invasive Argentine ant. Conservation Biology 14: 559563.

Hedges. S.A. 1998: Field guide for the management of structure-invading ants. Franzak and Foster, Cleveland, OH.

Holway, D.A. 1995: Distribution of the Argentine ant (Linepithema humile) in northern California. Conservation Biology 9: 16341637.

Holway, D.A. 1998a: Factors governing rate of invasion: a natural experiment using Argentine ants. Oecologia 115: 206212.

Holway, D.A. 1998b: Effect of Argentine ant invasions on ground-dwelling arthropods in northern California riparian woodlands. Oecologia 116: 252258.

Holway, D.A. 1999: Competitive mechanisms underlying the displacement of native ants by the invasive Argentine ant. Ecology 80: 238251.

Holway, D.A.; Suarez, A.V.; Case, T.J. 1998: Loss of intraspecific aggression in the success of a widespread invasive social insect. Science 282: 949952.

Science for Conservation 196 33 Hooper-Bui, L.M.; Rust, M.K. 2000: Oral toxicity of abamectin, boric acid, fipronil, and hydramethylnon to laboratory colonies of argentine ants (Hymenoptera: Formicidae). Journal of Economic Entomology 93: 858886.

Human, K.G.; Gordon, D.M. 1996: Exploitation and interference competition between the invasive Argentine ant, Linepithema humile, and native ant species. Oecologia 105: 405412.

Human, K.G.; Gordon, D.M. 1997: Effects of argentine ants on invertebrate biodiversity in Northern California. Conservation Biology 11: 12421248.

Human, K.G.; Weiss, S.; Weiss, A.; Sandler, B.; Gordon, D.M. 1998: Effects of abiotic factors on the distribution and activity of the invasive Argentine ant (Hymenoptera, Formicidae). Environmental Entomology 27: 822833.

Keller, L.; Passera, L. 1990: Fecundity of ant queens in relation to their age and the mode of colony founding. Insectes Sociaux 37: 116130.

Keller, L.; Passera, L.; Suzzoni, J.P. 1989: Queen execution in the Argentine ant, Iridomyrmex humilis. Physiological Entomology 14: 157163.

Kennedy, T.A. 1998: Patterns of an invasion by Argentine ants (Linepithema humile) in a riparian corridor and its effects on ant diversity. American Midland Naturalist 140: 343350.

Klotz, J.H.; Greenberg, L.; Venn, E.C. 1998: Liquid Boric acid bait for control of the Argentine ant (Hymenoptera, Formicidae). Journal of Economic Entomology 91: 910914.

Klotz, J.H.; Greenberg, L.; Amrhein, C.; Rust, M.K. 2000a: Toxicity and repellency of borate-sucrose water baits to Argentine ants (Hymenoptera: Formicidae). Journal of Economic Entomology 93: 12561258.

Klotz, J.; Greenberg, L.; Venn, G. 2000b: Evaluation of two hydramethylnon granular baits for control of Argentine ant (Hymenoptera : Formicidae). Sociobiology 36: 201207.

Knight, R.L.; Rust, M.K. 1991: Efficacy of formulated baits for control of Argentine ant (Hymenoptera: Formicidae). Journal of Economic Entomology 84: 510514.

Krushelnycky, P.D.; Reimer, N.J. 1998a: Bait preference by the Argentine ant (Hymenoptera : Formicidae) in Haleakala National Park, Hawaii. Environmental Entomology 27: 1482 1487.

Krushelnycky, P.D.; Reimer, N.J. 1998b: Efficacy of maxforce bait for control of the Argentine ant (Hymenoptera: Formicidae) in Haleakala National Park, Maui, Hawaii. Environmental Entomology 27: 14731481.

Leathwick, J.R.; Stephens, R.T.T. 1998: Climate surfaces for New Zealand. Unpublished Landcare Research Contract Report LC9798/126.

Liang, D.; Silverman, J. 2000: You are what you eat: Diet modifies cuticular hydrocarbons and nestmate recognition in the Argentine ant, Linepithema humile. Naturwissenschaften 87: 412416.

Majer, J.D. 1993: Spread of Argentine ants (Linepithema humile), with special reference to Western Australia. Pp. 163173 in Williams, D.F. (Ed.) Exotic ants; biology, impact, and control of introduced species. Westview Press. Boulder, CO.

Markin, G.P. 1970: Food distribution within laboratory colonies of the Argentine ant, Iridomyrmex humilis (Mayr). Insectes Sociaux 17: 127158.

Morrison, L.W. 1997: Polynesian ant (Hymenoptera, Formicidae) species richness and distribu- tiona regional survey. Acta OecologicaInternational Journal of Ecology 18: 685695.

Oi, D.H.; Williams, D.F.; Koehler, P.G.; Patterson, R.S. 1994: Imported fire ants and their management in Florida. University of Florida, Extension Service SP161. 20 p.

Orr, M.R.; Seike, S.H. 1998: Parasitoids deter foraging by Argentine ants (Linepithema humile) in their native habitat in Brazil. Oecologia 117: 420425.

Osborne, B. 1998: Argentine ant survey of the Mount Maunganui/Arataki area, Bay of Plenty. Unpublished report, Environment Bay of Plenty, Tauranga.

34 HarrisPotential impact of the Argentine ant in New Zealand Passera, L. 1993: Characteristics of tramp species. Pp. 2343 in Williams, D.F. (Ed.) Exotic ants: Biology, impact, and control of introduced species. Westview Press, Boulder. CO.

Phillips, P. 1986: Field trials show effective control of Argentine ants. California Grower 10: 1617.

Porter, S.D.; Alonso, L.E. 1999: Host specificity of fire ant decapitating flies (Diptera: Phoridae) in laboratory oviposition tests. Journal of Economic Entomology 92: 110114.

Porter, S.D.; Savignano, D.A. 1990: Invasion of polygyne fire ants decimates native ants and disrupts arthropod community. Ecology 71: 20952106.

Porter, S.D.; Van Eimeren, B.; Gilbert, L.E. 1988: Invasion of red imported fire ants (Hymenoptera: Formicidae): microgeography of competitive replacement. Annals of the Entomological Society of America 81: 913918.

Reimer, N.J. 1993: Distribution and impact of alien ants in vulnerable Hawaiian Ecosystems. Pp. 11 22 in Williams, D.F. (Ed.) Exotic ants: Biology, impact, and control of introduced species. Westview Press, Boulder, CO.

Reimer, N.J.; Cope, M.; Yasuda, G. 1993: Interference of Pheidole megacephala (Hymenoptera; Formicidae) with biological control of Coccus viridis (Homoptera: Coccidae) in coffee. Environmental Entomology 22: 483488.

Ripa, R.; Rodriguez, F.; Rust, M.K.; Larral, I; Robinson, W.H. 1999: Distribution of liquid food and bait in colonies of Argentine ant (Hymenoptera: Formicidae). Pp. 225229 in Rettich, F.; Rambo, G.W. (Eds) Proceedings of the 3rd International Conference on Urban Pests. Czech University of Agriculture, Prague, Czech Republic, 1922 July 1999.

Rupes, V.; Chmela, J.; Ledvinka, J. 1997: Comparison of the efficacy of baits with sulfluramid, hydramethylnon and methoprene against pharaohs ant. International Pest Control 39: 189191.

Sanders, N.J.; Barton, K.E.; Gordon, D.M. 2001: Long-term dynamics of the distribution of the invasive Argentine ant, Linepithema humile, and native ant taxa in northern California. Oecologia 127: 123130.

Shattuck, S.O. 1999: Australian ants: their biology and identification. Collingwood, Australia, CSIRO Publishing. Monographs on invertebrate taxonomy: vol. 3.

Sherley, G. 1996: Morphological variation in the shells of Placostylus species (Gastropoda, Bulimulidae) in New Zealand and implications for their conservation. New Zealand Journal of Zoology 23: 7382.

Shorey, H.H.; Gaston, L.K.; Gerber, R.G.; Sisk, C.B.; Phillips, P.A. 1996: Formulating farnesol and other ant-repellent semiochemicals for exclusion of Argentine ants (Hymenoptera, Formicidae) from citrus trees. Environmental Entomology 25: 114119.

Skarratt, D.B.; Sutherest, R.W.; Maywald, G.F. 1995: CLIMEX for windows version 1.0: Users guild. CRC for Tropical Pest Management, Brisbane.

Stimac, J.L.; Pereira, R.M.; Alves, S.B.; Wood, L.A. 1990: Field evaluation of a Brazilian strain of Beauveria bassiana for control of the red imported fire ant, Solenopsis invicta, in Florida. Page 337 in Proceedings and abstracts of the 5th international colloquium on invertebrate pathology and microbial control, Adelaide, Australia.

Stringer, C.E. jr; Lofgren, C.F.; Bartlett, F.J. 1964: Imported fire ant toxic bait studies: evaluation of toxicants. Journal of Economic Entomology 57: 941945.

Suarez, A.V.; Bolger, D.T.; Case, T.J. 1998: Effects of fragmentation and invasion on native ant communities in coastal southern California. Ecology 79: 20412056.

Suarez, A.V.; Richmond, J.Q.; Case T.J. 2000: Prey selection in horned lizards following the invasion of Argentine ants in southern California. Ecological Applications 10: 711725.

Suarez, A.V.; Holway, D.A.; Case, T.J. 2001: Patterns of spread in biological invasions dominated by long-distance jump dispersal: Insights from Argentine ants. Proceedings of the National Academy of Sciences of the United States of America 98: 10951100.

Science for Conservation 196 35 Tsutsui, N.D.; Suarez, A.V.; Holway, D.A.; Case, T.J. 2000: Reduced genetic variation and the success of an invasive species. Proceedings of the National Academy of Sciences of the United States of America 97: 59485953.

Valentine, E.W.; Walker, A.K. 1991: Annotated catalogue of New Zealand Hymenoptera. DSIR Plant Protection Report 4. General Printing Services, Nelson.

Van Schagen, J.J.; Davis, P.R.; Widmer, M.A. 1993: Ant pests of Western Australia, with particular reference to the Argentine ant (Linepithema humile). Pp. 174-180 in Williams, D.F. (Ed.) Exotic ants: Biology, impact, and control of introduced species. Westview Press, Boulder, CO.

Vega, S.J.; Rust, M.K. 2001: The Argentine ant: A significant invasive species in agricultural, urban and natural environments. Sociobiology 37: 325.

Vinson, S.B.; Sorensen, A.A. 1986: Imported fire ants life history and impact. Texas Department of Agriculture, Texas.

Visser, D.; Wright, M.G.; Giliomee, J.H. 1996: The effect of the Argentine ant, Linepithema humile (Mayr) (Hymenoptera, Formicidae), on flower-visiting insects of Protea nitida Mill (Proteaceae). African Entomology 4: 285287.

Ward, P.S. 1987: Distribution of the introduced Argentine ant (Iridomyrmex humilis) in natural habitats of the lower Sacramento valley and its effects on the indigenous ant fauna. Hilgardia 55: 116.

Way, M.J.; Cammell, M.E.; Paiva, M.R. 1992: Studies on egg predation by ants (Hymenoptera: Formicidae) especially on the eucalyptus borer Phoracantha semipunctata (Coleoptera: Cerambycidae) in Portugal. Bulletin of Entomological Research 82: 425432.

Way, M.J.; Cammell, M.E.; Paiva, M.R.; Collingwood, C.A. 1997: Distribution and dynamics of the Argentine ant Linepithema (Iridomyrmex) humile (Mayr) in relation to vegetation, soil conditions, topography and native competitor ants in Portugal. Insectes Sociaux 44: 415 433.

Wetterer, J.K. 1998: Nonindigenous ants associated with geothermal and human disturbance in Hawaii Volcanoes National Park. Pacific Science 52: 4050.

36 HarrisPotential impact of the Argentine ant in New Zealand