Board Meeting: September 2015 Agenda Number: 12 (i) File Ref: PU0024

LORD HOWE ISLAND BOARD

Business Paper ITEM

Lord Howe Island Rodent Eradication Program Update

BACKGROUND

On 18 May 2015, after the community consultation process over late 2014 and early 2015 ending with the community survey, the Board decided to proceed with the planning and approvals stage of the Program leading towards implementation of the rodent eradication plan, if the required approvals were received. The decision-making process was captured in a diagram, which assisted the Board on the day and which has been amended to include additional information and is attached. The rodent eradication program has now been divided into three stages: Stage One: Preliminary planning and community consultation

This stage has already been completed. It involved undertaking required initial trials including captive management and toxin resistance trials as well as initial operational planning. It included the biosecurity review and progressing of biodiversity outcome monitoring. Finally it included the community consultation and engagement process and the community survey.

Stage Two: Planning and Approvals

This stage is now underway. The key tasks during this stage are:

• Assemble personnel to undertake the work on the next stages • Review the Rodent Eradication Plan to ensure that it takes into consideration all new

information since it was drafted in 2009 • Develop individual property and livestock management plans, which will inform the

eradication plan and the approval process. This will involve a detailed property by property consultation with individual leaseholders and residents.

• Undertake any necessary studies required for the approval process, including independent health assessment

• Continue the relevant baseline outcome monitoring • Finalise detailed planning and all necessary risk assessments; • Obtain required permits and approvals, • Update and finalise operational details; • Prepare tender documentation

Stage Three: Implementation and evaluation of the eradication plan

This Stage will not happen until Stage Two is completed.

After all the necessary approvals are obtained and the required planning is undertaken, the decision-makers, that is the Commonwealth and State funding bodies and the Board will consider all the information and make the decision about proceeding to Stage Three.

Stage Three will involve the eradication plan being implemented in winter 2017 over a three month period. In summary the key tasks leading up to and during this stage are:

• Establish monitoring programs • Procurement of bait and helicopter operators • Recruitment of temporary field staff for hand baiting • Captive management of selected non target species • Carrying out the baiting operation • Environmental monitoring • Evaluation of the outcome of the eradication

CURRENT POSITION

1. Assembling the personnel and resources The Project Manager’s position was advertised because the existing Project Manager, Pete McClelland, was not able to return to the Island for personal reasons. However Pete will continue to be involved with the Program from off the Island in reviewing strategies and plans until the Project Manager position is filled, and then be available for advice and consultation after that. Interviews for the Project Manager position will be held in Sydney on 15 September. One of the Assistant Project Manager positions was advertised and recruitment has now been completed. The successful candidate is Anthony Wilson, currently Catchment Coordinator, Northeast Catchment Management Authority, Victoria. Anthony will be moving to the Island to take up the position with his wife and young son in early October. There is another Assistant Project Manager position in the Program structure. However recruitment to this position will not take place until the planning and approvals have been further progressed.

2. Project timelines

The Planning and Approvals Stage will take up to two years to complete. There is significant work to be undertaken in preparing and gathering information from property management and making applications for various approvals. The approvals also take time to be processed and determined.

Following is a broad outline of the steps in Stage 2: Planning and Approvals. It is subject to change depending on the time taken for each step.

Step Description Start Date End Date Status

1 Assemble resources 1 July 2015 30 June 2016

Engage staff to undertake the Planning and Approvals work:

- Project Manager - Asst Project Manager (Community) - Asst Project Manager (Operations)

Develop role descriptions for field staff

3 /8/15 1/715 1/3/16

31/10/15 30/9/15 30/6/16

2 Communications and consultation 1 July 2015 30 June 2017

Communications Plan developed Communications and consultation undertaken

3 Property Management Plans 1 Oct 2015 31 Mar 2016

Develop 130 Property Management Plans Develop 30 Livestock Management Plans

4 Captive Management arrangements 1 Nov 2015 30 June 16

Species to be managed – agreed Contract with Taronga Zoo drafted

1/11/15 1/1/16

31/12/15 30/6/16

5 Operations Plan update 1 Oct 2015 31 Mar 2016

Update to include operational constraints arising from Prop & Livestock Mgt Plans.

1/10/15 31/03/16

6 Permits and approvals 31 Mar 2016 30 Nov 2016

Prepare applications for approvals: • Species impact statement • APVMA approval • EPCB approval • Captive holding permit • Animal ethics approval • NSW Pesticides Act approval • POEO Act Flight Ops approval • CASA Flight Ops approval

7 Human health assessment review 31 Mar 2016 30 Nov 2016

Identify appropriate reviewer Request for review to be undertaken

8 Biodiversity outcome monitoring 1 Nov 2015 30 June 2017

Monitoring Plan confirmed Monitoring undertaken Report completed

9 Biosecurity arrangements 1 Nov 2015 30 June 2017

Finalise Biosecurity Plan Develop implementation plan Implement plan (subject to funding)

10 Preparation of contracts 30 June 2016 28 Feb 2017

Tender documentation prepared for: • Bait procurement • Helicopter operations • Captive management

11 Technical and financial feasibility assessment

31 Dec 2016 30 Mar 2017

The feasibility assessment will take into account any changes to the methodology resulting from the planning and approvals process and cost estimates for the implementation.

11 Stage 3 Implementation Plan 31 Dec 2016 30 June 17

Decision made by LHIB, CfOC and ET to proceed to Stage 3 Stage 3 Implementation Plan developed

3. Outcome Monitoring report

The monitoring project is being undertaken by the Office and Environment and Heritage’s Science Division under the project leadership of Dr Nicholas Carlile. It is an ongoing process taking place over the duration of the Rodent Eradication Program. For 2014/2015, there were numerous trips to the Island by project staff looking at different animal and plant species.

The following studies were undertaken during the 2014/2015 year:

• Terrestrial bird study • Seabird study of Black Noddy • Reptile study of the LHI Skink.

• Seabird study of Black-winged Petrel • Flora studies of Mountain Palms • Flora studies of other flower plants

A Final Report (attached) of this part of the Outcome Monitoring Project has been produced: Monitoring, evaluating and reporting the biodiversity benefits of eradicating exotic rodents from Lord Howe Island.

RECOMMENDATION

That the Board note the LHI Rodent Eradication Program update

Prepared __________________ Penny Holloway Chief Executive Officer

Rodent EradicationProcess for resolution

•2015 Community consultation outcome•Majority view to proceed•Level of support likely to impact success

•APVMA Approval process•EPBC Approval process•NSW approval processes•Independent Health assessment incorporating international evidence and advice

•Board Go/No go decision•Informed by:•Status of approvals•Additional advice from chief scientist•Ongoing community consultation

Principle 1:Board needs to respond to majority view

Principle 2: Board committed to attempting to resolve outstanding community concerns

Principle 3:Board will act to keep our people safe, and to protect the Island’s environment, World Heritage status and unique tourism assets

Ongoing Community Consultation

i

OFFICE OF ENVIRONMENT AND HERITAGE

Monitoring, evaluating and reporting the biodiversity benefits of eradicating exotic rodents from Lord Howe Island.

Final report:15 June 2015

Nicholas Carlile Science Manager, Ecosystem Management Science

ii

Contents

Summary iv

Background 1

Flora Studies Tony Auld, Andrew Denham, Mark Tozer, Ian Hutton 4

Introduction 4

Methods Impact of Black rats on fruits and seeds 6

Impacts of Black rats on the Little Mountain Palm on Mt Gower 8

Extinction risk for the Gnarled Mossy Cloud Forest EC 9

Results Impact of Black rats on fruits and seeds 9

Impacts of Black rats on the Little Mountain Palm on Mt Gower 12

Extinction risk for the Gnarled Mossy Cloud Forest EC 14

Discussion 15

Reptile Studies Robert Wheeler, George Madani 18

Introduction 18

Methods 19

November 2014 19

April 2015 22

Results and Discussion 22

North Bay November 2014 22

Salmon Beach November 2014 26

North Bay / Salmon Beach April 2015 27

Recommendations 29

Bird Studies 30

Seabird studies for identifying candidates for biodiversity benefits monitoring

on Lord Howe Island Nicholas Carlile, Lisa O’Neill, Ben Hope 30

Introduction 30

White Tern 31

Methods 31

Results and Discussion 32

Grey Ternlet 32

iii

Methods 32

Results and Discussion 34

Black Noddy 34

Methods 35

Results and Discussion 35

Black-winged Petrel 36

Methods 36

Results and Discussion 37

Seabird Recommendations 38

Land birds on Lord Howe Island Peter Fullagar, Chris Davey, A O Nicholls and

Ian Hutton SUMMARY 39

Land Bird Recommendations 39

Acknowledgements 40

References 41

Appendix 1 44

Appendix 2 (PDF) Assessing impacts of introduced rats on Lord Howe Island

plants including the Little Mountain Palms and its habitat on Mt Gower

(with images) to be attached 45

Appendix 3 (PDF) Auld TD & Leishman MR (2015). Ecosystem risk assessment for

Gnarled Mossy Cloud Forest, Lord Howe Island, Australia. Austral Ecology 40,

364–372 to be attached

iv

Summary

The Biodiversity Benefits Program associated with the rodent eradication project has

been established to assess and document the biodiversity benefits of removing rats

and mice from the World Heritage Lord Howe Island. The program provides a

measure of the return on investment. It also allows an evaluation of current status of

species and so any impacts of the eradication of rodents on key non-target species

can be tracked during their recovery. Over time, results from the various monitoring

components could be integrated to identify and explore changes to ecosystem

processes.

The following key species and taxa were targeted specifically during this revised two-

year project (2013-14 and 2014-15): Flora: Big Mountain Palm Hedyscepe

canterburyana; Little Mountain Palm Lepidorrhachis mooreana and fruiting plants

(including Blue Plum Chionanthus quadristamineus, Berrywood Ochrosia elliptica,

Green Plum Atractocarpus stipularis, Pandanus Pandanus forsteri and Tamana

Elaeodendron curtipendulum); Reptiles: LHI Skink Oligosoma lichenigera ; Seabirds:

White Tern Gygis alba; Grey Ternlet Procelsterna cerulean, Black Noddy Anous

minutus, Black-winged Petrel Pterodroma nigripennis; land birds.

Flora studies used selective caging and cafeteria trials to determine removal and

loss of fruits and seeds to Black Rats Rattus rattus. Of the 16 species examined,

losses to Black Rats were very high for six species (Howea forsteriana, Olea

paniculata, Baloghia inophylla, Jasminium simplicifolium, Smilax australis and

Geitonoplesium cymosum); potentially very high but variable for Berrywood;

moderate for three species (Syzygium fullagarii, Dietes robinsoniana and Blue Plum)

and low-moderate in the remaining 6 species. No species tested was entirely free of

seed or fruit losses to Black Rats. This suggests Black Rats are likely to impact on

many species across LHI and further testing of species is warranted to examine the

number of species likely to be at risk due to impacts on seed production, seed

survival and plant recruitment. In addition, further work on what impact the loss of

seeds or fruits may have on the affected plant populations is also necessary.

Additionally the high impact of Black Rats on the Little Mountain Palm on Mt Gower

that had been identified in 2007 was confirmed by repeat sampling in 2014. While

v

there is some successful recruitment in this species where rodent baiting occurs on

Mt Gower, recruitment failure is evident at locations over 100 m or so from bait

stations. This suggests that to successfully maintain this critically endangered palm

on Mt Gower rat control or eradication will be required across the Mt Gower summit.

The habitat of the Little Mountain Palm, the Gnarled Mossy Cloud Forest, was

assessed as critically endangered using the IUCN Red List for Ecosystems criteria.

Control of Black Rats on the summit of Mt Gower is the most practical means to

reduce the threat to this ecological community in the short term.

The LHI Skink is endemic to Lord Howe Island, and is listed as Vulnerable in both

NSW and national environmental legislation. Surveys for this species were

undertaken at various locations on the island in November 2014 and April 2015 to

obtain baseline data on the skink before an eradication takes place. Population

structure, relative abundance and distribution were attempted measurements for

comparative studies following the rodent eradication.

It was found that the skink is in low numbers, if present at all, in the southern and

central areas of Lord Howe Island. However, the population in North Bay appears to

be abundant although largely restricted to the sedge-grass plant community found on

sand dunes.

Surveillance camera studies on seabirds had mixed results. White Tern showed no

impact from rodents in incubation and chick provisioning periods but were present on

nesting branches when sites were inactive. Accessible Grey Ternlets nests were

scarce, reflecting the potential impacts of rodents on this species on the main island.

From two monitored nests, one was unsuccessful with subsequent rodent predation

of an unguarded egg. Black Noddy nesting was unaffected by rodents and none

were recorded at nest sites during the study. Black-winged Petrel were monitored

monthly during the breeding season and those nesting areas where rats were

uncontrolled had poor success in incubation and total failure of fledglings. In areas

where rats were controlled, breeding success was mixed but the results were also

confounded by heavy Masked Owl Tyto novaehollandiae predation of adults. If the

Biodiversity Benefits Program was to continue in its revised form future studies of

seabirds should concentrate on Grey Ternlet and Black-winged Petrel breeding sites

vi

using surveillance cameras. Consideration should also be given to monitoring Little

Shearwater Puffinus assimilis breeding sites using similar methods.

Comparing only two years of land bird surveys showed that most common bird was

the Golden Whistler Pachycephala pectoralis followed by the Silvereye Zosterops

lateralis. A noticeable reduction in all indices occurred for the next four most

frequently reported species – Blackbird Turdus merula, Pied Currawong Strepera

graculina, Magpie-lark Grallina cyanoleuca and Welcome Swallow Hirundo nigricans.

The suggested potential differences between 2013 and 2014 for some species (Buff-

banded Rail Gallirallus philippensis, Magpie-lark, Welcome Swallow and Blackbird)

need further exploration and additional observations before it can be determined that

the detected differences between years are truly significant.

1

Monitoring, evaluating and reporting the biodiversity benefits of eradicating exotic rodents from Lord Howe Island.

Final report: 15 June 2015

Background:

The Lord Howe Island group (LHIG; 31°31'S, 159°03’E) is located 760 kilometres

north east of Sydney. The group comprises the main island (Lord Howe Island; 1455

ha) and 28 smaller islets and rocks (Figure 1). The most significant of these are the

Admiralty Group (1 km to the north of LHI) and Balls Pyramid (23 km to the south

east). The island group is a World Heritage Area with global environmental

significance, and Australia has an international obligation to effectively manage the

island’s substantial, world-renowned natural values.

In common with a number of oceanic islands, populations of invasive rodents have

been accidentally introduced to Lord Howe Island (LHI). The house mouse Mus

musculus probably arrived before 1860, and the ship rat Rattus rattus in 1918. Both

species of introduced rodent have significant impacts on the biodiversity of Lord

Howe Island. Rats are implicated in the extinction of at least five endemic bird taxa

(species or subspecies), and the local extirpation of the endangered LHI Phasmid

Dryococelus australis, now found only on Balls Pyramid, and Wood-feeding

Cockroach Panesthia lata on offshore islands including the Admiralty Group. The

rodents are recognised as an ongoing threat to at least another 13 bird species, two

reptiles, 49 flowering plants, 12 vegetation communities, and numerous threatened

invertebrates, many of which are endemic to LHI. Rats also cause significant

damage to the Kentia Palm Howea forsteriana, with resulting economic losses to the

local palm industry. Predation by ship rats on LHI is listed as a Key Threatening

Process under both New South Wales (NSW) state and Australian Government

legislation.

The eradication of rodents from LHI will have enormous benefits for a range of

species currently adversely affected by rats and mice, including a suite of threatened

endemic species. In particular, we expect increases in the abundance of seeds and

2

seedlings, thereby enhancing the process of forest regeneration as well as increases

in the abundance and distribution of invertebrates, reptiles, seabirds, and land birds.

The monitoring program associated with the rodent eradication project is currently

assessing and documenting the biodiversity benefits of removing rats and mice. This

provides a measure of the return on investment. It also evaluates any impacts of the

eradication on key non-target species and will track their recovery in the absence of

rodents. Over time, results from the various monitoring components could be

integrated to identify and explore changes to ecosystem processes.

The initial project plan (April 2013 to June 2018) intended to research and monitor

key taxa identified in the Rodent Eradication Plan to determine the changes in the

distribution and abundance of those species. The work commenced prior to the

actual eradication operation planned at the time, to allow collection of baseline, pre-

eradication data, and was to continue for at least three years after the eradication to

assess the benefits of the operation. The following taxa were targeted specifically:

Kentia Palm; Big Mountain Palm; Little Mountain Palm and other fruiting plants; land

snails; the LHI Wood-feeding Cockroach; LHI Gecko and LHI Skink; LH Pied

Currawong; land birds and seabirds. In addition, data from the annual surveys of LH

Woodhen conducted by the LHI Board was to be incorporated into the integrated

analysis. By December 2013 only surveys on targeted flora, land birds and selected

seabirds had occurred as the progress of the eradication plan altered due to

changing circumstances on LHI.

A revised project plan was amended in March 2014 in response to the decision

taken in early 2014 to move away from immediate implementation of the eradication

program to full community engagement and consultation. In response and to further

minimise costs, the scope of the monitoring program was reduced from the original

eradication plan. Reviewing existing data, the project refocused on those species or

taxa for which existing monitoring data was absent or inadequate. It is envisaged

that full monitoring could commence once implementation of the eradication program

resumes, with 1-2 years of data collectable before baiting occurs.

3

The following key species and taxa being targeted specifically during this revised

two-year project (2014-15 and 2015-16): Flora: Big Mountain Palm Hedyscepe

canterburyana; Little Mountain Palm Lepidorrhachis mooreana and fruiting plants

(including Blue Plum Chionanthus quadristamineus, Berrywood Ochrosia elliptica,

Green Plum Atractocarpus stipularis, Pandanus Pandanus forsteri and Tamana

Elaeodendron curtipendulum). Reptiles: LHI Skink Oligosoma lichenigera; Seabirds:

Black-winged Petrel Pterodroma nigripennis, Black Noddy Anous minutus; and land

birds.

This final report covers the period August 2013 to May 2015 (Appendix 1) as agreed

to in the LHIB Revised Project Plan, March 2014.

4

Flora Studies

Assessing impacts of introduced rats on Lord Howe Island plants including the Little Mountain Palms and its habitat on Mt Gower

Tony D Auld1, Andrew Denham1, Mark Tozer1 and Ian Hutton2

1, NSW Office of Environment and Heritage, PO Box 1967 Hurstville NSW 2220

2, PO Box 157, Lord Howe Island, NSW 2898

Email: tony.auld@environment.nsw.gov.au

Introduction

The flora of Lord Howe Island is recognised as globally significant due to the high

level of endemism and unique vegetation communities. Auld and Hutton (2004)

detail this as: “The flora of Lord Howe Island has a high level of endemism and many

of the floristic assemblages are also unique to the island group (Pickard 1983, Green

1994). There are five plant genera endemic to Lord Howe Island — Negria,

Lordhowea and three palms, Howea, Hedyscepe and Lepidorrachis. For the vascular

plants, Green (1994) lists 459 species, 241 indigenous (53%), of which 105 are

endemic (44%) and 218 naturalised (48%). Of the indigenous vascular plants, 58

species are ferns and 183 are flowering plants. Several more species of both

indigenous and naturalised plants have been found since Green’s (1994) work. The

high level of endemism is typical of islands and comparable with megadiverse

regional areas of continents (Lowry 1998).”

It is widely recognised that invasive species may have significant negative impacts

on oceanic islands, particularly for many narrow range endemics and ecological

communities. While Lord Howe Island (LHI) was originally free of mammals (except

for small insectivorous bats) when first encountered by humans in 1788 and first

settled in 1834 (Hutton 1986), Black Rats Rattus rattus were accidentally introduced

when a steamship ran aground in 1918 (Billing and Harden 2000). A number of plant

5

species and vegetation communities have been identified as being at risk from the

impacts of Black Rats and House Mice Mus musclus on LHI. These impacts include:

• Loss of fruits and seeds to rats, e.g. Baloghia inophylla, Chionanthus

quadristamineus, Drypetes deplanchei, Elaeodendron curtipendulum,

Hedyscape canterburyana, Howea forsteriana, Howea belmoreana,

Lepidorrhachis mooreana, Pandanus forsteri, Ochrosia elliptica (Auld &

Hutton 2004, Auld et al. 2010);

• loss of seedlings and stem damage (the four palm species, Hedyscape

canterburyana, Howea forsteriana, Howea belmoreana, Lepidorrhachis

mooreana, Dietes robinsoniana (Wedding Lily) and at least two fern species,

Asplenium milnei and Adiantum hispidulum) (Auld & Hutton 2004);

• impacts on the critically endangered Gnarled Mossy Cloud Forest Ecological

Community confined to the summits of the southern mountains (NSW

Scientific Committee 2011);

As well, the extinctions of two LHI plant species, Sicyos australis and Solanum

bauerianum, were most likely influenced by consumption of seeds and fruits by

introduced animals (Auld & Hutton 2004; NSW Scientific Committee 2010). Solanum

bauerianum is now considered to be globally extinct.

In this work, we undertook to quantify the impact of introduced rats on 2 aspects of

the vegetation of LHI. Firstly, we wanted to understand the breadth of plants species

that may be impacted by rats. To do this, in the field we tested the magnitude of fruit

and seed loss to rats across a range of species, with the species tested being

dictated by fruit and seed availability. For some species we were able to replicate

the trials in different locations or at different times of the year. Secondly, we re-

examined the size distribution of plants of the Little Mountain Palm Lepidorrhachis

mooreana on Mt Gower. Previously, Auld et al. (2010) had shown that rats were

preventing the establishment of new plants into the parts of the population that were

not baited. We re-measured these sites to determine if the impact of rats was

ongoing and to try and see if rat baiting was leading to sufficient protection for these

palms.

6

Finally, through funding support from an Australian Research Council Grant to TA

and others we assessed the extinction risk for the Gnarled Mossy Cloud Forest

Ecological Community using the recently developed IUCN Red List for Ecosystems

criteria (Keith et al. 2013).

Methods Impact of Black Rats on fruits and seeds

At one or more sites (See Table 1) we established 5 plots beneath mature plants of

several species (Kentia Palm Howea forsteriana, Berrywood Ochrosia elliptica,

Maulwood Olea paniculata, Coprosma putida) (Table 1). Each plot contained three

treatments: uncaged; caged to exclude only birds (rodents could enter in gaps at

each end (see Fig. 1 Appendix 2); and caged to exclude both rodents and birds. At

each plot, in each treatment we placed 5 mature fruits. The density of fruits so

placed was not dissimilar to that found naturally below plants with mature fruits. We

recorded the fate of seeds over 3-7 days. We also established an infared camera at

one plot for each species to record what animal was responsible for seed removal or

loss (these cameras were set up to activate and take 10 photos when a warm

blooded animals comes into view of the camera (i.e. a rat , a mouse, a bird, a

human).

Table 1 Seed and fruit losses across species and sites. Family Scientific

Name Common Name

Life form

Experiment Type

Site Date of trial

Seeds lost to rats (%)

Fleshy Fruit species Arecaceae Howea

forsteriana Kentia Palm

Palm Single Neds Beach

Nov-13

100

Howea forsteriana

Single Research Station

Nov-13

100

Howea forsteriana

Single Windy Ridge W

Nov-13

100

Howea forsteriana

Single Windy Ridge

Nov-13

100

Howea forsteriana

Single Windy Ridge E

Nov-13

80

Howea forsteriana

Single Neds Beach

Jan-14

100

Oleaceae Olea paniculata Maulwood Tree Single Research Station

Nov-13

100

Olea paniculata Single Middle Beach

Nov-13

100

7

Olea paniculata Single Windy Ridge W

Nov-13

100

Olea paniculata Single Windy Ridge E

Nov-13

100

Apocynaceae Ochrosia elliptica

Red Berrywood

Shrub Single Neds Beach

Nov-13

96

Ochrosia elliptica

Single Middle Beach

Nov-13

96

Ochrosia elliptica

Single Windy Ridge W

Nov-13

12

Ochrosia elliptica

Single Windy Ridge E

Nov-13

4

Myrtaceae Syzygium fullagarii

Scalybark Tree Single Research Station

Jun-14

44

Syzygium fullagarii

Cafeteria Erskine Valley

Jun-14

40

Rubiaceae Coprosma putida

Stinkwood Shrub Single Research Station

Jun-14

0

Coprosma putida

Single Erskine Valley

Jun-14

16

Oleaceae Jasminium simplicifolium

Jasmine Vine Cafeteria Research Station

Nov-13

64

Jasminium simplicifolium

Cafeteria Middle Beach

Nov-13

100

Jasminium simplicifolium

Cafeteria Windy Ridge E

Nov-13

100

Jasminium simplicifolium

Cafeteria Windy Ridge

Nov-13

100

Smilacaceae Smilax australis Vine Cafeteria Research Station

Nov-13

92

Smilax australis Cafeteria Middle Beach

Nov-13

100

Smilax australis Cafeteria Windy Ridge E

Nov-13

92

Smilax australis Cafeteria Windy Ridge

Nov-13

96

Smilacaceae Geitonoplesium cymosum

Vine Cafeteria Research Station

Nov-13

68

Geitonoplesium cymosum

Cafeteria Middle Beach

Nov-13

100

Rutaceae Sarcomelicope simplicifolia

Yellow Wood

Tree Cafeteria Research Station

Nov-13

12

Rubiaceae Psychotria carronis

Black Grape

Tree Cafeteria Erskine Valley

Jun-14

20

Iridaceae Dietes robinsoniana

Wedding lily

Herb Cafeteria Erskine Valley

Jun-14

36

Oleaceae Chionanthus quadristamineus

Blue Plum Tree Cafeteria Erskine Valley

Jun-14

32

Meliacaea Dysoxylon pachyphyllum

Island Apple

Tree Cafeteria Erskine Valley

Jun-04

4

Old Fruit with no pulp Euphorbiaceae Drypetes

deplanchei Greybark Tree Cafeteria Research

Station Nov-13

8

Cafeteria Middle Beach

Nov-13

32

Dry seeds Fabaceae Sophora Lignum Tree Cafeteria Research Nov- 32

8

howinsula Vitae Station 13 Sophora

howinsula Cafeteria Middle

Beach Nov-13

36

Sophora howinsula

Cafeteria Windy Ridge E

Nov-13

0

Sophora howinsula

Cafeteria Windy Ridge

Nov-13

0

Euphorbiaceae Baloghia

inophylla Bloodwood Tree Cafeteria Neds

beach Nov-13

100

Baloghia inophylla

Cafeteria Middle Beach

Nov-13

100

Baloghia inophylla

Cafeteria Windy Ridge E

Nov-13

100

Baloghia inophylla

Cafeteria Windy Ridge

Nov-13

100

We also established plots for multiple species in a ‘cafeteria’ trial within the habitat

where they generally occur. We did this for 12 species at 5 sites (See Table 1). In

this trial 5 fruits or seeds of 4-6 species were placed together in small clumps (see

Fig. 2 Appendix 2). At each plot all seeds were uncaged. Again, like above, the fate

of the fruits/seeds was followed over 3-7 days if possible.

Impact of Black Rats on the Little Mountain Palm (Lepidorrhachis mooreana) on Mt

Gower

We repeated the sampling of Auld et al. (2010) (see end of Appendix 2) and sampled

the size structure of stands of Little Mountain Palms in cloud forest on the Mount

Gower plateau. We stratified the sampling between areas that have been baited by

the Lord Howe Island Board to control rat numbers and those that have never been

baited. In each of the baited and unbaited areas we sampled three separate plots for

the Little Mountain Palm (6 plots). We also sampled an additional 4 plots in the

gradient between baited and unbaited areas. We estimated the distance for each of

our 10 sampled plots from an existing bait station to determine if there was any

additional benefit of baiting into surrounding unbaited areas. At each plot, we

established a 5 m wide transect across the site and sampled all individual palms

within the transect. The transects were sampled until we had encountered a

minimum of 30-50 Little Mountain Palm individuals with an emergent trunk. We

9

measured the trunk height to the base of the leaves, the number of leaf scars for

individuals with a trunk and the height of individuals for individuals without a trunk.

Extinction risk for the Gnarled Mossy Cloud Forest Ecological Community

To assess the Gnarled Mossy Cloud Forest against the IUCN Red List for

Ecosystems criteria (Keith et al. 2013), we searched the available literature for

evidence of any threats to the ecosystem that may result in ecosystem decline. We

assessed the Gnarled Mossy Cloud Forest against four of the IUCN criteria (see

Appendix 2, Auld and Leishman 2015 for more details):

• Decline was assessed using historical vegetation reports (Maiden 1898, 1914;

Oliver 1916; Pickard 1983; Green 1994; Mueller-Dombois & Fosberg 1998;

Harris et al. 2005). For more recent changes, we inspected satellite imagery

(Google Earth and 2011 air photo imagery).

• For Restricted geographic distribution we used available Geographical

Information System data layers for the distribution of Gnarled Mossy Cloud

Forest (mapped from Pickard 1983) to estimate both the extent of occurrence

and area of occupancy of the ecosystem.

• To assess changes in abiotic features that directly impact on Gnarled Mossy

Cloud Forest, we used two metrics based on collated data from the Australian

Bureau of Meteorology from mid 1940s onwards (Rainfall and cloud cover).

• For biotic degradation, we reviewed the impact of exotic rats on limiting

recruitment of key species in the ecosystem and in changing the structure and

composition of the ecosystem.

Results Impact of Black Rats on fruits and seeds

In total, we were able to test the removal of fruits or seeds in 16 species, comprising

a mixture of plant families, life forms (trees, shrubs, vines) and habitats. Most

species examined had a fleshy fruit. In summary, seed or fruit losses were apparent

in all study species, at least at some sites.

Virtually all seed losses occurred at night and there was little removal of fleshy fruits

by birds from our ground based trials. It is likely that most bird removal of fruits

10

occurs on the plants and not the ground. Occasionally a shearwater returning to its

nest disturbed one of our plots and scattered fruits, but this was not recorded as fruit

loss in our calculations. Evidence of seed or fruit losses to rats was indicated by:

• Camera images of rats removing or consuming fruits/seeds (see Figs 3-6

Appendix 2)

• In situ seed or fruit remains indicating rats (eg, incisor marks etc) (see Figs 7-

11 Appendix 2)

We found no clear pattern of greater loss in areas without rat baiting stations present

versus those with baiting stations.

Trees

Howea forsteriana

There was high levels of seed removal in this species in all trials (5 sites, 2 time

periods, Table 1), with 5/6 trials with 100% fruit loss and 1/6 with 80% loss.

Olea paniculata

All four trails across 4 different sites showed 100% seed loss.

Syzygium fullagarii

Fruit loss was around 40-44% at 2 study sites. For both sites, the trials only ran for 2-

3 days so further fruit losses could be expected.

Sarcomelicope simplicifolia

12% fruit loss at one site (insufficient fruits available for further study).

Psychotria carronis

20% fruit loss at one site (trial only ran for 2 nights so further losses may occur).

Chionanthus quadristamineus

32% fruit loss at one site (trial only ran for 2 nights so further losses may occur).

Dysoxylon pachyphyllum

4% fruit loss at one site (trial only ran for 2 nights so further losses may occur).

11

Drypetes deplanchei

Seed loss ranged from 8-32% across 2 sites. Only fruits with no flesh available for

trials.

Sophora howinsula

Seed loss ranged from 0-36% across four sites. When eaten seed, seeds were

consumed in situ (see Fig. 8).

Baloghia inophylla

100% seed loss in all trials across 4 sites.

Shrubs

Ochrosia elliptica

Fruit loss varied from 4-96% across sites (Table 1). One of the sites with low losses

was in a location with no adult plants. The cameras showed clear evidence of rats

removing fruits (Fig. 4 Appendix 2). On other occasions fruits were eaten in situ and

fruit remains where present.

Coprosma putida

Fruit loss varied from 0-16% across 2 study sites.

Dietes robinsoniana

36% fruit loss at one site (trial only ran for 2 nights so further losses may occur).

Vines

Jasminium simplicifolium

Total fruit loss (100%) at three sites and 64% at the fourth

Smilax australis

Fruit loss varied from 92-100% across four sites.

Geitonoplesium cymosum

68-100% fruit loss across 2 sites.

12

Impact of Black Rats on the Little Mountain Palm (Lepidorrhachis mooreana) on Mt

Gower

Based on sampling in 2006, Auld et al. (2010) showed that not only were the fruits of

Little Mountain Palm consumed by rats, but the recruitment and establishment of a

juvenile seedling bank only occurred in the baited habitats on Mt Gower. Our re-

sampling of these and other sites in 2014 found the pattern remains unchanged. Two

of the 3 sites that were unbaited lacked new seedlings and small juveniles (eg. Fig.

12) and one of these showed a decline in larger juveniles over time, suggesting

ongoing recruitment failure. The unbaited site nearest the baiting stations showed a

small pulse of juveniles less than 50 cm in height, suggesting that that there may

have been some decline in rat impacts for a period in this area. The three sites in the

baited area, all showed evidence of some small juveniles being present, but the

abundance of small juveniles was greatly reduced in 2014 compared to 2006 (eg see

Fig. 13). For the 4 additional sites sampled, one had no juvenile plants at all, two had

no seedlings (one of these sites had only 1 juvenile less than 50 cm high), while only

one site had a few seedlings.

Fig. 12. A comparison of juvenile plants sizes between 2006 and 2014 at an unbaited site of Little Mountain Palm on Mt Gower. Note the lack of small juveniles and seedlings.

seedl

<50cm

<100cm

<150cm

<200cm

>200cm

Juvenile plant size (cm)

0

2

4

6

8

10

12

Freq

uenc

y

20142006

YEAR

13

Fig. 13. A comparison of juvenile plants sizes between 2006 and 2014 at a baited site of Little Mountain Palm on Mt Gower. Note the presence of small juveniles and seedlings (particularly in 2006).

There was a pattern of greater recruitment of juvenile plants closer to bait stations

(Fig. 14) with recruitment of small plants declining to zero at plots sampled around

250 m from bait stations.

Fig. 14. Abundance of seedlings (black circles) and small juveniles (black triangles) of Little Mountain Palms in relation to distance from nearest bait station on Mt Gower.

seedl

<50cm

<100cm

<150cm

<200cm

>200cm

Juvenile plant size (cm)

0

100

200

300

400

500

Freq

uenc

y

20142006

YEAR

0 100 200 300 400 500Distance to nearest bait station (m)

0

10

20

30

40

50

60

70

Abun

danc

e

0 100 200 300 400 500Distance to nearest bait station (m)

0

50

100

150

Abun

danc

e

14

Extinction risk for the Gnarled Mossy Cloud Forest Ecological Community

Overall the Gnarled Mossy Cloud Forest is considered to be Critically Endangered

based on a restricted geographic distribution combined with continuing decline.

Historical Decline: There have been no observed changes in the extent of this

ecosystem in the last 50 years (Pickard 1983). No changes are indicated since

Pickard’s 1983 vegetation classification based on inspection of recent satellite

imagery (Google Earth and 2011 air photo imagery) and field traverses. The status

of the ecosystem is therefore Least Concern under criterion A1.

Geographical Extent and Continuing Decline: The geographic distribution of the

ecosystem is below the IUCN threshold for Critically Endangered (Auld and

Leishman 2015). Decline was inferred from: a loss of moisture from declining rainfall

and cloud cover due to climate change (affecting disturbance regimes, gap formation

and species survival and recruitment); ongoing exotic rat predation on seeds and

seedlings of several sensitive species that are structural components of the

ecosystem (affecting survival and recruitment); and the fact that the ecosystem is

considered to exist at only one location (Auld and Leishman 2015).

Decline in abiotic processes and features: There was a trend for decline in both

rainfall and the frequency of cloud cover although there is still a great deal of

variation in the degree of decline predicted (Auld and Leishman 2015, see end of

Appendix 3).

Decline in biotic processes and interactions: Rats damage key functional plant

species in the ecosystem via predation on seeds and seedlings (Auld & Hutton 2004;

Auld et al. 2010). Most of the area (93%) of Gnarled Mossy Cloud Forest is subject

to impact by rats (7% is baited to reduce impact). The extent and severity of seed

predation by rats is high for two key species (mountain palms, but the severity of

seed predation by rats on other taxa in the ecosystem is uncertain. More work on

this aspect is required.

15

Discussion Impact of Black Rats on fruits and seeds

There was evidence that rats were consuming fruits or seeds in all 16 species

examined. In summary, seed or fruit losses were apparent in all study species, at

least at some sites. Losses were very high for six study species (Howea forsteriana,

Olea paniculata, Baloghia inophylla, Jasminium simplicifolium, Smilax australis and

Geitonoplesium cymosum); potentially very high but variable for one species

(Ochrosia elliptica); moderate for three species (Syzygium fullagarii, Chionanthus

quadristamineus, Dietes robinsoniana) (the actual losses may be higher as the trials

only ran for a short period); generally low in 4 species (Sarcomelicope simplicifolia,

Psychotria carronis, Dysoxylon pachyphyllum, Coprosma putida) (but the actual

losses may be higher where the trials only ran for a short period); and low-moderate

in two species (Sophora howinsula, Drypetes deplanchei).

Further work on examining the impact of fruit losses on the ecology of the study

species would assist interpretation of these data. Does the impact of these losses

result in reduced potential for recruitment of new plants in the study species as has

been previously demonstrated for the Little Mountain Palm and Big Mountain Palm

Hedyscepe canterburyana (Auld et al. 2010, Simmons et al. 2012)? Given that

losses are occurring in all tested species, it would also be worthwhile to extend the

sampling to additional species as many other species are also likely to be impacted

by rats.

Impact of Black Rats on the Little Mountain Palm (Lepidorrhachis mooreana) on Mt

Gower

Repeat sampling in 2014 confirmed the earlier findings of Auld et al. (2010) that

Black Rats are significantly impacting on Little Mountain Palms on Mt Gower. Baiting

appears to be effective in allowing the development of a juvenile bank of palms,

although there was evidence that the size of this juvenile bank was greatly reduced

in 2014 compared to 2006. Sites within about 100m of bait stations had some

seedlings and small juveniles (indicating recent recruitment), but sites 250m or

further away mostly had zero recent recruitment. This suggests that to successfully

16

maintain this critically endangered palm on Mt Gower rat control or eradication

across the Mt Gower summit will be required.

Extinction risk for the Gnarled Mossy Cloud Forest Ecological Community

Assessment of the Gnarled Mossy Cloud Forest against the IUCN Red List for

Ecosystems criteria indicated that the ecosystem is critically endangered due to its

very highly restricted geographic distribution (confined to summits of Mt Gower and

Mt Lidgbird) combined with continuing decline due to a loss of moisture from

declining rainfall and cloud cover as a result of climate change (affecting disturbance

regimes, gap formation and species survival and recruitment); ongoing exotic rat

predation on seeds and seedlings of several sensitive species that are structural

components of the ecosystem (affecting survival and recruitment); and the fact that

the ecosystem is considered to exist at only one location (see Auld and Leishman

2015, Appendix 3 for details. This supports the current listing of this ecological

community as critically endangered under the NSW Threatened Species

Conservation Act (NSW Scientific Committee 2011).

The recognition that the introduced Black Rat is a threat to the Gnarled Mossy Cloud

Forest mirrors similar threats from exotic species and climate change to other Pacific

island cloud forests (Meyer et al. 2010). Eradication of rats from Lord Howe Island

will reduce the immediate risk to this ecosystem, however, only global mitigation of

greenhouse gases could alleviate risk from declining cloud cover and moisture

availability. If rat eradication does not occur or is unsuccessful, regular rodent baiting

across much of the distribution of the ecosystem, particularly on Mt Gower would be

needed to alleviate the threat. This may not be a long-term solution as the rodents

may gain resistance to the poisons. A reduction in rat impact is needed to both allow

successful seed production and recruitment of new plants in the Gnarled Mossy

Cloud Forest.

17

Reptile Studies

Lord Howe Island Skink Survey: November 2014 and April 2015

Robert Wheeler1, George Madani2

1, NSW Office of Environment and Heritage, PO Box 1967 Hurstville NSW 2220

2, chiro_ptera@hotmail.com

Email: robert.wheeler@environment.nsw.gov.au

Introduction

Funding has been obtained from the New South Wales and Australian governments

to undertake the eradication of introduced rodents from Lord Howe Island.

Monitoring key non-target species is crucial to evaluating the biodiversity benefits of

such an eradication. One such species is the Lord Howe Island Skink Oligosoma

lichenigera which is listed as Vulnerable under both the NSW Threatened Species

Conservation Act 1995 and the Commonwealth’s Environment Protection and

Biodiversity Conservation Act 1999.

This species is restricted to Norfolk Island and Lord Howe Island (DECC 2007)

although Cogger et al. (2005) suggest that the two island populations are genetically

distinct, and should not be regarded as the same species. On the Lord Howe group

of islands the skink is present on the main island, Balls Pyramid, Blackburn Island

and Roach Island (DECC 2007) as well as many smaller islets (Carlile & Priddel

2013a; 2013b; 2013c; 2013d). A wide range of vegetation communities, ranging from

lowland rainforest to montane forest as well as grasslands on the islets appear to be

acceptable to the skink (DECC 2007). On the main island, the skink seems to be

confined to sedge-grass habitat (Bray personal communication), the dense structure

of which may protect the skink from predators.

The skink feeds on beetles, spiders, ants and other invertebrates amongst the leaf

litter (DECC 2007) and fruit (Bray pers. com.).

18

Rats prey upon this species and are probably the main reason for its decline on the

main island (DECC 2007). The introduced Delicate Skink Lampropholis delicata,

which arrived in the early 1990’s, has spread from the settlement to the Northern

Hills and Intermediate Hill, and may compete for food with this species (DECC

2007).

To obtain baseline information on the status of the Lord Howe Island Skink (LHI

Skink) prior to rodent eradication, this survey of the lizard was undertaken to

determine:

• its relative abundance and distribution;

• population demography; and

• the extent of dispersal into habitats fringing sedge-grass habitat, currently the

stronghold of the species.

In addition, the feasibility of using tracking tunnels to assess broader distribution

patterns was evaluated.

Method

November 2014

Two potential study sites, North Bay and Windy Point, were inspected as to

suitability following information supplied by Rebecca Bray, a PhD candidate studying

Lord Howe Island’s two endemic lizard species, the LHI Skink and the Lord Howe

Island Gecko Christinus guentheri. The sedge-grass community located on the

foredune of North Bay is known to contain the largest population of the LHI Skink

(Bray, unpublished field-trip report September 2010). It may be no coincidence that

North Bay is also regularly baited for rats to protect the Lord Howe Island Flax Snail

Placostylus bivaricosus.

In November 2014, Sooty Terns Onychoprion fuscata were nesting along the dune,

particularly in the northern and central sections. The southern section had a

relatively small number of tern nests so it was selected as one study site as this

would result in little disturbance to these birds. This area corresponds to Rebecca

Bray’s North Bay Dune site. The sedge-grass community at Windy Point, which is

19

approximately 3 km south-east of the North Bay site, was known to be the next-most

populated site for the skink (Bray, unpublished field-trip report September 2010).

Unfortunately, approximately half of this site has since been washed away. The

sedge-grass community at Salmon Beach at the southern end of Lord Howe Island

was chosen as a substitute for Windy Point.

At each site, two lines of pitfall traps (20 traps at each site) were set in the sedge-

grass community on the sand dunes, and ran approximately parallel to the nearby

seashore. Distance between traps in each line was 5 metres. At North Bay the lines

were identified as Green (traps 1-12) and Blue (traps 1-8). The sedge-grass

community trap lines at Salmon Beach were separated by 7 metres, and both

contain ten traps each, with 5 metres between sequential traps. Additionally at each

site, pitfall traps were set in the adjacent habitats. For North Bay, two lines (the North

and Middle lines) ran at 3250 to the sedge-grass community lines, and passed

through a shrub community dominated by hopbush Dodonaea viscosa burmanniana

then into a Kentia Palm Howea forsteriana forest. The North and Middle lines each

had one trap in the sedge-grass community, one on the boundary of sedge-grass

community and the adjacent habitat and eight traps in the adjacent communities. A

South Line (one trap in the sedge-grass, one on the boundary and four in adjacent

habitat) ran through a Kentia Palm forest growing on shallow soil over limestone with

rock outcrops. At Salmon Beach a line of ten traps ran perpendicular to the sedge-

grass community through lowland rainforest towards Mt Lidgbird. Traps in this

Lidgbird Line were left closed for two days after being set as a precaution in case the

smell of newly turned soil attracted rodents or the Lord Howe Woodhen Gallirallus

sylvestris to them.

Nearest-neighbour distance for traps 1, 2, 3 and 4 in the lines into adjacent habitat

(i.e., the North, Middle, South and Lidgbird lines) was 5 metres. Except for the South

Line, distance between traps after trap 4 was 10 metres. For the South Line traps,

the nearest-neighbour distance for all the traps was 5 metres, and, unlike the straight

line followed in the North, Middle and Lidgbird lines, the course of the South Line

was somewhat zig-zagged because of the scarcity of soil deep enough to fully burry

the pit-trap.

20

Each pit-trap line consisted of a series of 5-litre buckets (190 mm deep, 210 mm

across, MaxiPailR) buried into the soil up to the rim. A quarter of a pear (for bait)

(Goulburn ValleyR tinned pears), several leaves (for shelter) and pieces of pumice or

polystyrene (for floatation in case of flooding) were placed into each bucket. A

minimum of eight holes, each 8 mm in diameter, was also drilled into the bottom of

each bucket for drainage. However two traps on the South Line (and situated on a

slope) filled with water as a consequence of moderate rainfall during one morning.

Traps were set for four consecutive days and nights at the dune sites (North Bay and

Salmon Beach) and the adjoining habitat sites at North Bay, in the period 18 – 26

November 2014. Traps were left open overnight with the exception of those traps

along the Lidgbird line, which were closed at night to safeguard against being raided

by rats. The pears used to lure skinks into the pit-traps were missing from the

majority of traps at the Lidgbird site by late afternoon on the second day of trapping

which suggested that this line was also being visited by currawongs or woodhen,

hence the Lidgbird line was closed after only two days.

Traps were cleared between 7:30 and 10 a.m., and again around mid-day and late

afternoon/dusk. Skinks caught were weighed using a Pesola spring balance,

measured (snout-vent) on a ruler, and individually numbered using a non-toxic

marker pen (Sharpie Fine-point Permanent MarkerR). Sex determination was partly

based on colouration. A yellowish or copper-colour underside was taken as evidence

of the skink being male; gravid females had pronounced bulges to their sides.

Differentiation between adults and sub adults was based on the amount of colouring

(bright colouration is indicative of breeding status) and animal size. Only data

collected from those skinks caught in the sedge-grass community, the known

stronghold of the species, has been used to examine demography.

To evaluate the practicality of using tracker tunnels for surveying skink distribution,

two captured skinks were released at one end of a tunnel, and encouraged to leave

at the other end, and in so doing, pass over an area painted with printer’s ink. The

aim was to determine if the resulting foot prints were sufficiently clear to distinguish

tracks made by the LHI Skink from that of the Delicate Skink.

21

April 2015

Surveys for the skink were conducted on Lord Howe Island between the 16 - 26 April

2015 at the two sites examined in November 2014 as well as an additional site at the

playground where a survey had been undertaken by Rebecca Bray in April-May

2011. The additional site was included due to the lack of capture of the LHI Skink at

Salmon Beach the previous spring.

Trap lines opened at North Bay (along the dune, and into the adjacent habitat) and at

Salmon Beach (along the dune) were the same as for November but the line leading

towards Mount Lidgbird from Salmon Beach was not used due to the significant

presence of predators. Two lines, each of 10 traps, were set out approximately

parallel to the beach near the Playground, and another line of 10 traps ran

perpendicular to these, and passed through a Norfolk Pine-Kentia Palm community.

Nearest-neighbour spacing of traps at this site was 5 metres.

Traps were baited with pieces of pear, which was replenished after three days.

Trapping was conducted at Salmon Beach and the playground from 16 – 26 April,

and at North Bay from 20 – 26 April. Trapping was not conducted on the 21st and

22nd April when over 230 mm of rain was recorded falling on Lord Howe Island.

Results and Discussion

North Bay November 2014

In the equivalent of 92 pit-trap days at the sedge-grass dune site at North Bay (12

traps in the Green Line, eight in the Blue Line, plus the first trap of the North, Middle

and South lines, each line opened for four days and nights) 123 LHI Skinks were

caught, with daily totals of 34, 25, 37, and 27. There were only three recaptures. This

contrasts markedly with the trapping result for September 2010 when 32 LHI Skinks

were caught in 20 traps over five days (100 pit-trap days) in the same area (Bray,

unpublished field-trip report September 2010). In September 2010, recaptures

numbered four, a 12.5% recapture rate compared to less than 3% for November

2014. It is possible that the ink marking used to number the skinks was lost due to

abrasive contact between the marked lizard and the coarse sedge vegetation or

sand, and this resulted in recaptured individuals being recorded as new captures by

22

mistake. When descriptions of lizards of the same length (+ 1mm) are compared in

regard to noted features (such as sex, weight, tail characteristics, when and where

caught) there were at least 99 individuals caught at North Bay in November 2014 so

it is possible that there may have been up to 24 other recaptures that were recorded

as new captures. Although, the time between the initial and subsequent capture with

the three known recaptures was no more than two days, the marking was still

distinctive so it is likely that a substantial proportion of these 24 captures were of

new animals. Even if it is accepted that there were only 99 individuals caught, and

not 123, this still represents a three-fold increase in the September 2010 total with

approximately the same capture effort. The difference may be due to a rise in the

skink population or, perhaps, to an increase in skink activity (and, therefore,

probability of capture) as spring progressed.

Sixty-three of the dune skinks (25 females, 24 males, 14 sex unknown; 34 adults and

29 sub-adults) were noted to have wounds to some degree, ranging from scratch

marks on the head and body to a regenerated tail or part thereof. Such wounds may

be a product of robust mating activity or possibly antagonistic behaviour as a result

of overcrowding.

Of the 67 captured LHI Skinks from North Bay that were judged to be adult, at least

42 are female but only 19 to 25 are believed to be male (Table 1) although at least

40% of the sub-adult population is comprised of males. It is unlikely that the unequal

sex ratio for the adults is a product of trapping bias with adult males being somehow

less prone to falling into traps. The approximately two-to-one bias in favour of

females may represent the adult population structure of the dune colony. More than

half the number of skinks caught in, or at the border with, the adjacent habitats were

believed to be male (6 males, 5 unknown sex; Table 2). More than 81% of the skinks

caught in the nearby habitats or at the very edge of the sedge-grass community were

judged to be sub-adults. These findings suggest that males, particularly young

males, are being disproportionably excluded from the better habitat.

Four skinks were caught in the Dodonaea shrub community, at 5 metres (n = 2, one

of which was an adult), 10 metres (n = 1) and 20 metres (n = 1) from the sedge

grass area. Three, including one adult, were caught in the palm forest up to 70

23

metres from the dune, and one skink was caught in the rock-outcrop area (i.e., along

the South Line) 15 metres from the sedge-grass community.

No Delicate Skinks were caught at North Bay although they were commonly seen

along the path on the eastern side of the Dawson Point Ridge.

Table 1: Size and gender information for those Lord Howe Island Skinks caught in the sedge-grass

dune community at North Bay in November 2014 (Table 1a) and April 2015 (Table 1b). For the

determination of sex and age, see text. Length, in millimetres, the distance between the tip of the

snout and the vent; weight in grams.

Table 1a: Capture data from November 2014

All

captured

skinks

Sub-

adults

sex

unknown

Adults

sex

unknown

Sub-

adults

female

Adults

female

Sub-

adults

male

Adults

male

Number

caught

123 33 7 1 41 22 19

Mean

length

(SD; n)

63.0

(8.1; 122)

55.4

(3.7; 32)

63.9

(6.6; 7)

58 69.2

(4.9; 41)

56.6

(3.3; 22)

70

(5.8; 19)

Length

range

47-79 47-63 57-77 - 60-78 49-63 60-79

Mean

weight

(SD; n)

5.4

(2.2; 121)

3.6

(0.8; 33)

5.2

(1.2; 7)

4.2 7.1

(1.8; 40)

3.7

(0.5; 21)

6.9

(1.8; 19)

Weight

range

2.0-11.5 2.0-6.0 4.0-7.5 - 4.5-11.5 2.5-4.7 4.0-11

24

Table 1b: Capture data from April 2015

All

captured

skinks

Sub-

adults

sex

unknown

Adults

sex

unknown

Sub-

adults

female

Adults

female

Sub-

adults

male

Adults

male

Number

caught

32 2 2 9 4 15

Mean

length

(SD; n)

63.2

(9.1; 32)

50

(2.8; 2)

55

(1.4; 2)

67.7

(7.7; 9)

50

(3.8; 4)

66.3

(7.2; 15)

Length

range

47-76 48-52 54-56 58-76 47-56 56-76

Mean

weight

(SD; n)

5.63

(2.5; 26)

2.9

(0; 2)

3.45

(0.2; 2)

7.0

(2.4; 9)

2.6

(1.0; 3)

6.1

(2.5; 10)

Weight

range

2.2-9.7 2.9-2.9 3.3-3.6 4.4-9.7 2.2-4.2 3.2-9.1

25

Table 2: Size and gender information for those Lord Howe Island Skinks caught in habitats in the

vicinity of the sedge-grass dune community at North Bay in November 2014. For the determination of

sex and age, see text. Length, in millimetres, the distance between the tip of the snout and the vent;

weight in grams.

All

captured

skinks

Sub-

adults

sex

unknown

Adults

sex

unknown

Sub-

adults

female

Adults

female

Sub-

adults

male

Adults

male

Number

caught

11 5 0 0 0 4 2

Mean

length

(SD; n)

52.1

(8.2; 11)

49.6

(7.7; 5)

- - - 50.3

(6.9; 4)

62

(1.0; 2)

Length

range

39-63 39-59 - - - 39-58 61-63

Mean

weight

(SD; n)

3.2

(1.1; 11)

2.7

(1.0; 5)

- - - 3.1

(0.9; 4)

4.4

(0.4; 2)

Weight

range

1.5-4.8 1.5-4.0 - - - 2.0-4.5 4.0-4.8

Salmon Beach November 2014

The only skink species caught at Salmon Beach was the introduced Delicate Skink;

ten were caught in the sedge-grass community in 80 trap days, and one 70 metres

into the forest towards Mount Lidgbird during 20 trap days. In the sedge-grass

community, four skinks were caught on the first day, three on the second, two on the

third and one was caught on the last day; there were no recaptures.

Suitability of Tracking Tunnels

26

Two LHI Skinks were put through a tracking tunnel to evaluate the usefulness of

these tunnels for monitoring lizard distribution. The results were not encouraging.

The prints of a large skink (snout-vent length of 72 mm and weight of 8 g) were not

discernible being nothing more than smudges of ink on the running board, while a

smaller skink (snout-vent 52 mm, weight 3 g) got stuck on the ink board although it

eventually managed to free itself.

North Bay April 2015

Unlike in November, the greater majority of adult LHI Skinks caught at North Bay in

April are believed to be male (at least 53% compared to 28% for November) (Table

1). Overall, there were 35 individual skinks caught at this site, 32 in the dune traps

and three (an adult female, an adult male and a sub-adult of indeterminate sex) in

the adjacent habitat.

Sixty-seven LHI Skinks were trapped at North Bay over seven days/nights in autumn

2011 (Bray, unpublished field-trip report April-May 2011). The difference in the

numbers caught in 2011 compared to 2015 (67 and 32 respectively) may simply be

due to a difference in trapping effort, and not represent a decline in the population at

North Bay. Traps were closed for several days in April 2015 when recorded rainfall

exceeded 230 mm. Traps were also only opened during the day because of the

potential threat of overnight rain, therefore the trapping effort in 2015 was less than

that for 2011.

Three LHI Skinks were caught in the Dodonaea community up to 20 metres from the

sedge habitat. One Delicate Skink was also captured in this area. No LHI Skink was

caught in the palm forest in contrast to the capture result in November 2014.

Salmon Beach April 2015

The only skink found at Salmon Beach was the Delicate Skink; six captures being

recorded.

27

The playground April 2015

Low numbers of the LHI Skink were recorded at the playground in 2010 (e.g., three,

Bray unpublished fieldtrip report: Lord Howe Island Lizards - Nov-Dec 2010). No LHI

Skink was caught at the sedge-grass community or in the adjacent Norfolk Pine-

Kentia Palm habitat although five Delicate Skinks were trapped in the sedge-grass

community. One pit-trap was found to be without its pear bait on the sixth morning of

the survey which may indicate that the trap had been visited recently by a rodent or

currawong.

Current Status of the Lord Howe Island Skink

Trapping results from 2010 and 2011 (Bray, unpublished reports) and this survey

(2014 and 2015) indicate that the LHI Skink is, at best, present in very low numbers

in the southern and central areas of Lord Howe Island. The skink is, however,

abundant in the sedge-grass community growing on the dunes at North Bay. The

population also appears to be healthy; at least 44% of the adult females caught in

North Bay were carrying young in November. In addition, over 20% of the LHI Skinks

caught in 2014, and 18% of those caught in 2015 had been marked with translucent

tags by Rebecca Bray in 2010 and 2011 which may suggest a significant level of

survival for individuals as well as a high reproductive output.

Large numbers of LHI Skinks at North Bay may be a consequence of the rat baiting

that takes place in the area to protect the LHI Snail. However, the carcass of one rat

and the odour from several other dead rats were detected at the site in November

just after the vicinity was baited, so rats are still in the area. Baiting also takes place

at the playground (one bait station was located within two metres of a pit-trap in the

Norfolk Pine-Kentia Palm community) yet no LHI Skinks were found there, and a rat

was seen scurrying through the sedge-grass community at dusk. Mice also prey on

reptiles and their eggs, and can severely reduce reptile populations (Towns and

Broome 2003). If mice are not being controlled by the current baiting programme for

rats then it is likely that mice are also impacting the the populations of the LHI Skink.

Although the population at North Bay appears to be substantial and healthy, its long-

term future is not necessarily guaranteed. The species was once found in a variety of

habitats on the main island (DECC 2007) but now it is largely restricted to the sedge-

28

grass community found on sand dunes, and even here its distribution has shrunk to

the northern part of the island.

Recommendations

It is recommended that:

• rodent baiting continues at North Bay;

• the artificial wall built to stabilise the area at Windy Point is searched for the

LHI Skink;

• tracker tunnels are set in the sedge-grass community at the playground and

North Bay to determine the degree that rats or mice use these areas;

• reptile surveys are undertaken in the other habitats at North Bay if rodents are

eradicated from the main island; and

• the LHI Board considers translocating LHI Skinks to other parts of the main

island if a rodent-eradication programme proves successful.

29

Bird Studies

Seabird studies for identifying candidates for biodiversity benefits monitoring on Lord Howe Island

Nicholas Carlile, Lisa O’Neill, Ben Hope

NSW Office of Environment and Heritage, PO Box 1967 Hurstville NSW 2220

Email: nicholas.carlile@environment.nsw.gov.au

Introduction

The Biodiversity Benefits Program associated with the rodent eradication project was

established to assess and document the likely positive outcomes from the removal of

rats and mice. With costly programs for the eradication of rodents, it is desirable to

provide a meaningful measure of the return on investment. Despite the presence of

rodents on Lord Howe Island (LHI) for a century there has been little direct evidence

collected of their impact on seabirds. This group of animals makes up the most

significant component of the islands fauna group through the variety of species and

the sheer numbers of individuals. The potential extent of their occurrence over the

island, their annual breeding and foraging habits significantly add to, and possibly

‘drive’, the islands productivity through the transfer of oceanic nutrients to the

terrestrial system. The loss of species from the main island ecosystem and the

suppression of populations that breed here due to rodent impacts has the potential to

be catastrophic for the long-term viability of the current ecosystems that have

evolved here when they were more populous.

Two species of seabirds that once bred on the main island no longer occur here.

Populations of White-bellied Storm-petrel Fregetta grallaria thrive on the outer

islands (i.e. Carlile et al. 2013; Carlile & Priddel 2013a; 2013b;) and Kermadec Petrel

Pterodroma neglecta also occurs on the outer islands (Hutton 1991; Carlile & Priddel

2013a) but both were originally known from the southern mountains as well as other

locations on the main island (Hindwood 1940). It is likely, with the removal of rodents

that these species will return to breed on the main island.

30

Evaluation of any impacts of the eradication on key non-target species would be

most profitable monitoring changes in breeding response during recovery of species

currently breeding on the main island.

The revised Biodiversity Benefits Program specifically targeted spring-summer

breeding White Tern Gygis alba; Grey Ternlet Procelsterna cerulean, Black Noddy

Anous minutus and Black-winged Petrel Pterodroma nigripennis. It was considered

that these species were most likely to have suffer suppression of breeding success

(all species) and extent of occurrence (Grey Ternlet) since rodents, and specifically

Black Rats Rattus rattus became established on the island.

White tern

The White Tern breeds only within the settled lowlands, and colonies are highly

visible. The first substantiated breeding of this species on LHI was in 1968 (McAllen

et. al 2004). The preferred habitat is large, mature Norfolk Island Pines Araucaria

heterophylla that line Lagoon Road along the western shoreline of the island. These

trees were planted during the 1900s and have historical significance. Younger

stands of Norfolk Island Pines have since established within areas of native

vegetation. Concerned about the invasive nature of this species, the LHI Board has

declared Norfolk Island Pines an exotic weed (Lord Howe Island Board 2006), and

self-seeded stands that do not have historical value are being systematically

removed.

White Terns do not make a nest but lay their egg directly on a tree branch, usually

using a dimple or depression in the branch to help hold the egg on the branch. A

single egg is laid, though the terns do have capacity to lay a second and sometimes

even a third egg if prior eggs are lost.

Methods

Motion sensor cameras were set to monitor nests within the settlement area,

principally in the private garden of Wilsons Bike Hire business. Cameras were set

between 1.5 and 5.5 metres above the ground, attached to tree branches. In total 26

cameras were used to monitor 26 nests from egg laying (or just prior to laying) to

31

time of chick fledging or prior loss of the chick or egg. The majority of cameras were

set prior to nesting commencing. Over 629,000 images were captured at these nests

from 15th of September 2013 to 5th of April 2014 for analysis. Cameras were serviced

monthly to remove and download data cards, replace batteries and alter viewing

area or re-set camera positions depending on the outcome of the breeding attempt

at the monitored site.

Results and Discussion

The monitoring showed an extremely low survival rate from egg to fledgling chick.

Only three monitored chicks fledged and these chicks were not hatched in these

nests but were put into these nests as a young chick by ‘Macca’ of Wilsons Bike Hire

after being rescued from having fallen out of a nest elsewhere.

Of 25 first eggs laid, nine hatched (Hatching Success (HS) = 36%). There were 14

nests where a second egg was laid, ten of these were laid after loss of the previous

egg and four after loss of the previous chick. Of these 14 nests, eight eggs hatched

(HS = 57%). There were five nests where a third egg was laid. Three of these were

nests where the second egg was lost and two where the chick was lost. Of the five

third laid eggs, two hatched (HS = 40%). One nest was monitored from part way

through the season and it is not known if it was a first or subsequent nest, so is not

included in the analysis above. This nest successfully hatched its chick which was

later lost to an unknown cause.

In total, from 26 nests, 45 eggs were laid, 20 chicks hatched from those eggs, and

none survived to fledge. The causes of egg lost were varied, including falling from

the nest (4), predation by LH Currawong Strepera graculina crissalis (5) and egg

infertility (4). Of the remaining eggs (12 or 48%) that were lost, a cause of loss could

not be determined.

The major cause of chick loss was predation by LH Currawong, but one third of chick

disappearances could not be attributed to any cause. Chicks ranging in age from one

day old to 26 days old were predated by currawongs. Breeding Success (BS) at this

site was 0% although limited numbers of near-fledged young and flying juveniles

were observed at other locations (Carlile unpub data).

32

The lack of complete image capture from nest sites to determine egg and chick

losses highlights the failure of the infrared trigger system in the cameras to pick up

all bird movements. Birds are covered in feathers, which provide excellent insulation,

and do not allow indication of heat differentials between their body heat and the

ambient temperature. Only when the areas of high temperature differential, the front

of the head (eyes and open beak) or the underwings (when exposed while landing or

taking off) are presented to the infrared sensor does the camera trigger. Conversely,

mammals have an excellent thermal signal and would have been detected by the

cameras, so they can be excluded as the cause of the unknown nest failures.

White Tern showed no impact from rodents in incubation and chick provisioning

periods but Black Rats were present on nesting branches when sites were inactive

either before nesting or directly after depredation events. This indicates that the

presence of adult White Tern on nesting branches is a sufficient deterrent for rat

investigation of potential prey. Far from being threatened by either rodents or

predatory birds, the White Tern population continues to expand and is currently

estimated at 600 pairs in 2014, up from approximately 500 pairs in 2006 (Carlile &

Priddel 2015). Additional bird species observed during surveillance camera

monitoring of White Terns included two sightings of Long-tailed Cuckoo Eudynamys

taitensis 25/11/2013 and 14/12/2013.

Grey Ternlet

Grey Ternlets are a cavity nesting seabird, one of only two such species on LHI. The

White-bellied Storm-petrel Fregetta grallaria now only persists in nesting cavities on

the outer islands (i.e. Carlile et al. 2013) since the arrival of the Black Rat in 1918.

The Grey Ternlet, present on LHI since European occupation, is now confined to

occasional nesting sites within the North Hills and on the main islands east coast. It

is far more prevalent on the outer islands (Hutton pers comm), possibly as a result of

rodent interactions (McAllen et al. 2004).

Methods

Between the 16th and 21st of October 2013, surveys were made of the cliffs in the

North Hills in search of nesting Grey Ternlets which were expected to be incubating

33

or with freshly hatched young (Hutton 1991). Ian Hutton assisted in surveys around

Old Gulch and Dean Hiscox directed surveyors to sites at New Gulch. Surveillance

cameras were installed on accessible nests and monitored during the breeding

period until the fledging period in January 2014.

Results and Discussion

Surveys of Old Gulch recorded only two nests in total. The only accessible nest

already had a 10-day old chick in it and was deemed too advanced for camera

monitoring. Surveys at New Gulch recorded a total of six active Grey Ternlet nests

sites. Only two of these were accessible without the need for rope assistance, for

camera monitoring. One of the two nests had adults visiting at the time of camera

installation but the egg was cold, possibly indicating an abandoned nesting attempt.

Subsequent monitoring here showed that the egg was not being incubated and it

was removed by a Black Rat 10 days after the camera was installed. The second

nest was successful in raising a chick to fledging, and there were no images of rat

visitation. The chick was not left alone, even after initial brooding as adults remained

in attendance at the nest entrance at night for the duration of the in-nest provisioning

period as indicated by regular incursion to the nesting chamber during the hours of

darkness.

It is possible that by mid-October many nesting attempts had failed and the adult

Grey Ternlets had already abandoned their breeding sites. Future surveys for this

species should commence in early September and include Hells Gate and cliff lines

along the east coast of the main island as far as Red Point. Boat surveys in early

September would determine the most productive areas to search from the land for

camera monitoring.

Black Noddy

The Black Noddy is a tree nesting seabird, one of only two such species on LHI. It

has been breeding on LHI Island since 1989 (Hutton 1991). The population has, until

recently, been confined to North Bay where it has been expanding slowly in the

introduced Norfolk Island Pines behind the fore dune. Recently breeding had

commenced in a few Norfolk Island Pines along Lagoon Road in the settlement area

34

(Hank Bower pers comm). It was suspected that, if the population was breeding

successfully, a more rapid expansion would have been expected over the last 45

years. It is possible that the species is suffering from some form of reduced breeding

success that has hindered its expansion.

The Black Noddy lays a single egg in a shaped nest, usually on the branch of a large

established tree. There is some evidence to suggest that if the egg is lost a pair will

re-lay (this study).

Methods

A pilot study into the breeding success and causes of breeding failure for Black

Noddy was initiated at the start of October 2014. Initially, nine motion sensor

cameras were set up to monitor Black Noddy nests in introduced Norfolk Island

Pines at North Bay. Cameras were set at between 2.5 and 3.5 metres above the

ground. At this time a small number of nests were already with incubating eggs but

nest-building was still underway. An additional camera was added to make ten

cameras in November 2014. From these 10 cameras it was possible to monitor 12

nests.

Results and Discussion

Over 380,000 images of breeding activities were captured for analysis. Egg survival

to hatching was high, with only 2 eggs being lost, one fallen from the nest and one

cracked. A second egg was laid in the nest from which an egg fell, giving a total of

13 eggs able to be monitored from 12 nests. The replacement egg was successfully

incubated providing an overall nest HS of 91%. All pairs that successfully hatched

their chick (11), also raised it to fledging giving an overall BS of 91%.

There was no evidence on camera that rats were visiting or preying upon Black

Noddy at their nests. There was no evidence of any predation at any monitored nest.

The identical hatch and fledging success would indicate that the adult attendance

was high, preventing rodent access to young chicks. This was corroborated by the

surveillance cameras which showed adults on or at the nest during hours of

darkness. The study area at North Bay is included in the baiting program, principally

to protect the Lord Howe Island Flax Snail Placostylus bivaricosus. The low rodent

population may assist the Black Noddy in their high breeding success. The current

35

population at North Bay is 510 nesting pairs. If the effectiveness of baiting for

controlling rodents at this site declines due to rodent tolerance to current available

baits, or if baiting ceased to be carried out, Black Noddy breeding may be impacted.

This would warrant further investigation as to rodent impacts in any future

Biodiversity Benefits Program.

Black-winged Petrel

Black-winged Petrels are a burrow-nesting seabird, one of only two small burrowing

seabirds present on the main island of LHI. The Black-winged Petrel was first

observed breeding on LHI in 1971 (Hutton & Priddel 2002). The population has

established in several locations close to the coast at the northern end of the island

and along the eastern coast of LHI (McAllen et al. 2004), expanding slowly to an

approximate estimate of 500 pairs (Carlile unpub. data). It is suspected that if the

population was breeding successfully that a more rapid expansion would have been

expected over the last 45 years. On Phillip Island, off Norfolk Island, a similar

population established at the same time as the LHI birds (Priddel et al. 2003). Here,

they have expanded to between 5,000 and 10,000 pairs on an island without

introduced rodents (Carlile unpub. data). It is possible that the species on LHI is

suffering from some form of reduced breeding success that has hindered its

expansion.

Methods

Two sites were selected for a survey of breeding success of Black-winged Petrel,

with 21 burrows monitored on North Head and 35 burrows monitored just north of

Blinkys Beach on the main islands east coast. The North Head site is isolated from

the nearest rodent baiting grid by 200 m and is within lowland mixed forest. The

Blinkys Beach site is within a regularly maintained baiting grid of coastal forest

dominated by Melaleuca howeana and intersected by a fishers track to the eastern

shore of the island. The two sites were monitored roughly monthly (6-7 January, 10-

11 February, 9-10 March, 28-29 April) and their contents recorded. The burrows

were initially inspected to coincide with when the majority of eggs would have been

laid (Hutton & Priddel 2002) using an infrared burrow scope (Peep-A-Roo: Sandpiper

36

Technologies, Manteca, CA, USA). Following observations were timed to hatching

and early brooding of young by parents, half-grown pullus without parental protection

during daylight hours and near fledged young when abandoned by adults prior to

their fledging period (Hutton & Priddel 2002).

Results and Discussion

Success differed at the two sites. At North Head, two of 21 eggs (10%) were

observed to have hatched young in February 2015 and a single egg was still being

incubated. A further three adults were within their nesting chamber, possibly

incubating eggs (total: 4 potential eggs or maximum HS of 29% if all eggs survived to

this point). The sitting adults could not be shifted to emphatically indicate an egg

beneath them but their behaviour during disturbance by the burrowscope indicated

they were likely incubating an egg. There were six scatters of egg shell fragments

found on the surface with signs of rat chewing. It was not possible to determine

whether rats had destroyed the eggs while being incubated or after the eggs were

abandoned from failed nesting attempts. By March there was no sign of any eggs or

chicks and most nests were already deserted. In April, there were no adults in any of

the previously active nests indicating a BS of 0%.

At the Blinkys Beach site, 23 of 35 eggs (66%) were being incubated and a single

chick had hatched in February. By March there were seven chicks (HS = 20%)

observed and by April, five large chicks remained (BS = 14%). One of these chicks

was malformed and appeared unlikely ever to fly, reducing potential BS to 11%. The

nest site that had contained the freshly hatched chick in February was empty

indicating an additional chick may have fledged successfully but this could not be

confirmed. There were five adult bodies found in the lower, more open, section of the

Blinkys Beach colony near the fishers track that appeared to have been killed by

Masked Owl Tyto novaehollandiae.

The two sites showed differences in timing of laying, with proportionally more chicks

hatching early at North Head, but with much poorer egg survival at North Head. A

more thorough monitoring of nests, using surveillance cameras, may be able to

determine whether predation is a factor in the poor egg survival and hatching to

fledging survival once adults no longer attend the chicks (from about 3-5 days old) at

37

North Head. Cameras at the Blinkys Beach site may help determine the extent of owl

and rat depredation.

Seabird Recommendations:

• If the Biodiversity Benefits Program was to continue in its revised form future studies of seabirds should concentrate on Grey Ternlet and Black-winged Petrel breeding sites using surveillance cameras.

• Consideration should also be given to monitoring Little Shearwater Puffinus assimilis breeding sites using similar methods. This species is known to breed in both areas of rodent control (Priddel et al. 2003) as well as at other locations around LHI (Carlile pers comm). It has been established that this species shares burrows with summer-breeding Black-winged Petrel (Hutton & Priddel 2002) and, due to its similar breeding demography, would be a candidate for rodent predation, similar to that concluded from this Black-winged Petrel study.

38

Landbirds on Lord Howe Island, NSW

Survey report 2 (summary)

Note: Final reports have been provided elsewhere to the LHI Board

Peter Fullagar1, Chris Davey1, A O Nicholls1 and Ian Hutton2

1, Canberra Ornithological Group PO Box 301 CIVIC SQUARE ACT 2608

2, PO Box 157, Lord Howe Island, NSW 2898

Email: peter.fullagar@gmail.com

The purpose of the terrestrial bird study was to provide baseline information on the

numbers and distribution of land birds on Lord Howe Island. Point counts within

natural vegetation were chosen as the preferred method for the island survey. The

data collecting protocol used was one that best fitted the limitations for time and

effort available in the field. Additional transect counts within the settlement were

used to track the terrestrial bird presence in highly disturbed environments.

The 2013 and the 2014 survey covered 96 sites and were sampled four times within

the survey periods, distributed across the lowlands, by the Canberra Ornithologists

Group (Fullagar et al. 2014; Fullagar et al. 2015). Full results and site selection

methods are available in reports already with the Lord Howe Island Board.

In essence, when comparing only two years of land bird surveys, the most common

bird was the Golden Whistler Pachycephala pectoralis followed by the Silvereye

Zosterops lateralis. A noticeable reduction in all indices occurred for the next four

most frequently reported species – Blackbird Turdus merula, Pied Currawong

Strepera graculina, Magpie-lark Grallina cyanoleuca and Welcome Swallow Hirundo

nigricans. The suggested potential differences between 2013 and 2014 for some

species (Buff-banded Rail Gallirallus philippensis, Magpie-lark, Welcome Swallow

and Blackbird) need further exploration.

Land Bird Recommendations:

• Additional observations are required before it can be determined that the detected differences between years are truly significant.

39

Acknowledgements

Many field programs are reliant on the efforts of volunteers and this Biodiversity

Benefits Program is no different. We would like to acknowledge the efforts of Mark

Allen, Lia Battisson, David Cook, Kathy Cook, John Goldie, Christine Ledger, Martyn

Moffat, Gail Newmann, Michael Robbins and Kathy Walter from the Canberra

Ornithologists Group who assisted in the land bird surveys. Juliet Dingle, Margie

Grant, Ian Knight, Anicee Lombal, Melinda Norton and George Madani are thanked

for their assistance in seabird and reptile data collection. Lord Howe Island Board

assisted with transport, logistics and advice, particularly Christo Haselden, Hank

Bower and Michael ‘Fitzy’ Fitzgerald. Lord Howe Island residents Ian Hutton

provided support in the flora, land bird and seabird studies; Dean Hiscox provided

direction for seabird survey sites; Craig ‘Macca’ Campbell provided logistical support

and access to breeding seabirds on his property and we thank him and his family for

their patience, the paper is coming. The residents of the island shared their stories of

seabird experiences as well as their general knowledge of their habits and for that

we are grateful.

The assessment of extinction risk for the Gnarled Mossy Cloud Forest was

supported by an Australian Research Council Linkage Grant to TA and others (LP

LP130100435) and we thank Prof. David Keith and all the participants of the

Australian Centre for Ecological Analysis and Synthesis workshop for their thoughtful

contributions to the development of the risk assessment for this ecosystem

The project was funded by the Lord Howe Island Board under the Rodent

Eradication Program. Ethics approval was provided through the OEH animal ethics

committee numbers 021028/02 and 021028/01.

40

References

Auld TD & Hutton I (2004). Conservation issues for the vascular flora of Lord Howe Island. Cunninghamia 8, 490–500.

Auld T. D., Hutton I., Ooi M. K. J. & Denham A. J. (2010). Disruption of recruitment in

two endemic palms on Lord Howe Island by invasive rats. Biological Invasions 12, 3351-3361.

Auld TD & Leishman MR (2015). Ecosystem risk assessment for Gnarled Mossy

Cloud Forest, Lord Howe Island, Australia. Austral Ecology 40, 364–372. Billing J, & Harden B (2000). Control of introduced Rattus rattus L. on Lord Howe

Island. I. The response of mouse populations to warfarin bait used to control rats. Wildl Res 27, 655–658.

Carlile, N., Priddel, D. (2013a). Seabird islands No. 261: Mutton Bird Island, Lord

Howe Group, New South Wales. Corella 37 (4) 94-96. Carlile, N., Priddel, D. (2013b). Seabird islands No. 258: South Island, Lord Howe

Group, New South Wales. Corella 37 (4) 88-89. Carlile, N., Priddel, D. (2013c). Seabird islands No. 259: Sugarloaf Island, Lord

Howe Group, New South Wales. Corella 37 (4) 90-91. Carlile, N., Priddel, D. (2013d). Seabird islands No. 257: Tenth of June Island, Lord

Howe Group, New South Wales. Corella 37 (4) 86-87. Carlile, N. & Priddel, D. (2015) Establishment and growth of the White Tern Gygis

alba population on Lord Howe Island, Australia. Marine Ornithology 43: 113–118.

Carlile, N., Priddel, D., & Bower, H. (2013). Seabird islands No. 256: Roach Island,

Lord Howe Group, New South Wales. Corella 37 (4) 82-85. Cogger, H., Cameron, E. & Sadlier, R. (1979). The terrestrial reptiles of islands in the

Norfolk Island complex. Australian Museum, Sydney. pp.1—122. DECC (2007). Lord Howe Island Biodiversity Management Plan. Department of

Environment and Climate Change. Sydney. http://www.environment.gov.au/resources/parks/LHI_bmp.pdf

Fullagar, P., Davey, C., Nicholls, A. O. & Hutton I. Landbirds on Lord Howe Island,

NSW: Survey report 1, September 2013. Report prepared for the Lord Howe Island Board by the Canberra Ornithologists Group, February 2014.

Fullagar, P., Davey, C., Nicholls, A. O. & Hutton I. Landbirds on Lord Howe Island,

NSW: Survey report 2, September 2014. Report prepared for the Lord Howe Island Board by the Canberra Ornithologists Group, March 2015.

41

Green PS (1994). Flora of Australia Vol. 49 Oceanic Islands 1. Australian Government Printing Service, Canberra.

Harris R, Cassis G, Auld T & Hutton I (2005). Floristics and structure of the mossy

cloud forest of Mt Gower summit, Lord Howe Island. Pac. Conserv. Biol. 11, 246–56.

Hindwood, K. A. (1940). The birds of Lord Howe Island. Emu 40 1-86. Hutton, I (1986). Lord Howe Island (Conservation Press: Canberra ACT). Hutton, I. (1991). ‘Birds of Lord Howe Island: Past and Present’. (Ian Hutton: Coffs

Harbour). Hutton, I. & Priddel, D. (2002). Breeding biology of the Black-winged Petrel,

Pterodroma nigripennis, on Lord Howe Island. Emu 102 361-365. Keith DA, Rodríguez JP, Rodríguez-Clark KM et al. (2013). Scientific Foundations for

an IUCN Red List of Ecosystems. PLoS ONE 8, e62111. Lord Howe Island Board. 2006. Weed management strategy for Lord Howe Island. Lowry, PP (1998). Diversity, endemism, and extinction in the flora of New Caledonia:

a review. In Rare, threatened and endangered floras of Asia and the Pacific Rim (eds. C.-I Peng & P.P. Lowry II), (Institute of Botany, Academia Sinicia Monograph Series No. 16. Taipei). pp. 181–206.

Maiden JH (1898). Observations of the vegetation of Lord Howe Island. Proc. Linn.

Soc. N. S.W. 23, 112–58. Maiden JH (1914). Further notes on the botany of Lord Howe Island (fifth paper).

Proc. Linn. Soc. N. S.W. 39, 377–84. Meyer J-Y (2010). Montane cloud forests on remote islands of Oceania: the example

of French Polynesia (South Pacific Ocean). In: Tropical Montane Cloud Forests: Science for Conservation and Management (eds L. A. Bruijnzeel, F. N. Scatena & L. S. Hamilton) pp. 121–9. Cambridge University Press, Cambridge.

McAllan, I.A.W., Curtis, B.R., Hutton, I & Cooper R.M. (2004). The birds of the Lord

Howe Island Group: A review of records. Australian Field Ornithology supplement. 21, 1–82.

Mueller-Dombois D & Fosberg FR (1998). Vegetation of the Tropical Pacific Islands.

Springer-Verlag, NewYork. NSW Scientific Committee (2010). Final determination. Solanum bauerianum -

species presumed extinct listing [Cited 9th June 2015] Available from URL: http://www.environment.nsw.gov.au/determinations/solanumbauerianumFD.htm

42

NSW Scientific Committee (2011). Final determination. Gnarled Mossy Cloud Forest

on Lord Howe Island – critically endangered ecological community listing. [Cited 23September 2014.] Available from URL: http://www.environment.nsw.gov.au/determinations/gnarledmossycloudforestlhifd.htm

Oliver WRB (1916). The vegetation and flora of Lord Howe Island. Trans. Proc. N. Z.

Inst. 49, 94–161. Pickard J (1983). Vegetation of Lord Howe Island. Cunninghamia 1, 133–265. Priddel, D., Carlile, N., Evans, O., Evans, B & McCoy H. (2010). A review of the

seabirds of Phillip Island in the Norfolk Island Group. Notornis 57, 113-137. Priddel, D., Hutton, I., Carlile, N & Bester, A. (2003). Little Shearwaters, Puffinus

assimilis assimilis, breeding on Lord Howe Island. Emu 103 67-70. Simmons L, Auld TD, Hutton I, Baker WJ, & Shapcott A (2012). Will climate change,

genetic and demographic variation or rat predation pose the greatest risk for persistence of an altitudinally distributed island endemic. Biology 1(3), 736-765.

Towns, D.R. & Broome, K. G. (2003). From small Maria to massive Campbell: forty

years of rat eradications from New Zealand islands. New Zealand Journal of Zoology 30, 377-398.

Wilkinson IS & Priddel D (2011). Rodent eradication on Lord Howe Island:

challenges posed by people, livestock, and threatened endemics. In: Island Invasives: Eradication and Management (eds CR Veitch, MN Clout & DR Towns) pp. 508–14. IUCN, Gland.

43

Appendix 1

Timetable for the biodiversity studies on LHI August 2013 to May 2015.

Taxa

Rep

tiles

Sea

bird

s

Land

bird

s

Pal

ms

and

Frui

ting

plan

ts

Taxa

Rep

tiles

Sea

bird

s

Land

bird

s

Pal

ms

and

Frui

ting

plan

ts

Personnel

per trip 2 2 8 2

Personnel

per trip 2 2 8 2

2013-14 2014-15

July July

August August

September 5-16 8-14 6-12 September 10-17

October 16-21 October 6-10

November 15-19 18-24 November 17-27 9-13 17-23

December 12-18 December 16-20

January 14-17 January 5-9

February 15-18 February 9-13

March 7-10 March 16-26 9-13

April 04-07 April 27-01

May May

June 16-21 June Cancelled

1

Assessing impacts of introduced rats on Lord Howe Island plants including

the Little Mountain Palms and its habitat on Mt Gower

June 2015

Tony D Auld1, Andrew Denham1, Mark Tozer1 and Ian Hutton2

1, NSW Office of Environment and Heritage, PO Box 1967 Hurstville NSW 2220

2, PO Box 157, Lord Howe Island, NSW 2898

Email: tony.auld@environment.nsw.gov.au

Summary

We used selective caging and cafeteria trials to determine removal and loss of fruits

and seeds to Black Rats (Rattus rattus) on Lord Howe Island. Of the 16 species

examined, losses to Black Rats were very high for six species (Howea forsteriana,

Olea paniculata, Baloghia inophylla, Jasminium simplicifolium, Smilax australis and

Geitonoplesium cymosum); potentially very high but variable for one species

(Ochrosia elliptica); moderate for three species (Syzygium fullagarii, Chionanthus

quadristamineus, Dietes robinsoniana) (the actual losses may be higher as the trials

only ran for a short period) and low-moderate in the remaining 6 species. No species

tested was entirely free of seed or fruit losses to Black Rats. This suggests Black

Rats are likely to impact on many species across LHI and further testing of species is

warranted to examine the number of species likely to be at risk due to impacts on

seed production, seed survival and plant recruitment. In addition, further work on

what impact the loss of seeds or fruits may have on the affected plant populations is

also necessary.

The high impact of Black Rats on the Little Mountain Palm (Lepidorrhachis

mooreana) on Mt Gower that had been identified by Auld et al. (2010) was confirmed

by repeat sampling 8 years after the original sampling was done. While there is

some successful recruitment in this species where rodent baiting occurs on Mt

Gower, recruitment failure is evident at locations over 100 m or so from bait stations.

This suggests that to successfully maintain this critically endangered palm on Mt

Gower rat control or eradication will be required across the Mt Gower summit. The

habitat of the Little Mountain Palm, the Gnarled Mossy Cloud Forest, was assessed

as critically endangered using the IUCN Red List for Ecosystems criteria. Control of

Black Rats on the summit of Mt Gower is the most practical means to reduce the

threat to this ecological community in the short term.

2

Introduction

The flora of Lord Howe Island is recognised as globally significant due to the high level of endemism and unique vegetation communities. Auld and Hutton (2004) detail this as: “The flora of Lord Howe Island has a high level of endemism and many of the floristic assemblages are also unique to the island group (Pickard 1983, Green 1994). There are five plant genera endemic to Lord Howe Island — Negria, Lordhowea and three palms, Howea, Hedyscepe and Lepidorrachis. For the vascular plants, Green (1994) lists 459 species, 241 indigenous (53%), of which 105 are endemic (44%) and 218 naturalised (48%). Of the indigenous vascular plants, 58 species are ferns and 183 are flowering plants. Several more species of both indigenous and naturalised plants have been found since Green’s (1994) work. The high level of endemism is typical of islands and comparable with megadiverse regional areas of continents (Lowry 1998).” It is widely recognised that invasive species may have significant negative impacts on oceanic islands, particularly for many narrow range endemics and ecological communities. While Lord Howe Island was originally free of mammals (except for small insectivorous bats) when first encountered by humans in 1788 and first settled in 1834 (Hutton 1986), Black rats (Rattus rattus) were accidentally introduced when a steamship ran aground in 1918 (Billing and Harden 2000). A number of plant species and vegetation communities have been identified as being at risk from the impacts of Black Rats and house mice on LHI. These impacts include:

• Loss of fruits and seeds to rats, e.g. Baloghia inophylla, Chionanthus quadristamineus, Drypetes deplanchei, Elaeodendron curtipendulum, Hedyscape canterburyana, Howea forsteriana, Howea belmoreana, Lepidorrhachis mooreana, Pandanus forsteri, Ochrosia elliptica (Auld & Hutton 2004, Auld et al. 2010);

• loss of seedlings and stem damage (the four palm species, Hedyscape canterburyana, Howea forsteriana, Howea belmoreana, Lepidorrhachis mooreana, Dietes robinsoniana (Wedding Lily) and at least two fern species, Asplenium milnei and Adiantum hispidulum) (Auld & Hutton 2004);

• impacts on the critically endangered Gnarled Mossy Cloud Forest Ecological Community confined to the summits of the southern mountains (NSW Scientific Committee 2011);

As well, the extinctions of two Lord Howe Island plant species, Sicyos australis and Solanum bauerianum, were most likely influenced by consumption of seeds and fruits by introduced animals (Auld & Hutton 2004; NSW Scientific Committee 2010). Solanum bauerianum is now considered to be globally extinct. In this work, we undertook to quantify the impact of introduced rats on 2 aspects of the vegetation of LHI. Firstly, we wanted to understand the breadth of plants species that may be impacted by rats. To do this, in the field we tested the magnitude of fruit and seed loss to rats across a range of species, with the species tested being dictated by fruit and seed availability. For some species we were able to replicate the trials in different locations or at different times of the year. Secondly, we re-examined the size distribution of plants of the little mountain palm (Lepidorrhachis mooreana) on Mt Gower. Previously, Auld et al. (2010) had shown that rats were preventing the establishment of new plants into the parts of the population that were

3

not baited. We re-measured these sites to determine if the impact of rats was ongoing and to try and see if rat baiting was leading to sufficient protection for these palms. Finally, through funding support from an Australian Research Council Grant to TA and others we assessed the extinction risk for the Gnarled Mossy Cloud Forest Ecological Community using the recently developed IUCN Red List for Ecosystems criteria (Keith et al. 2013). Methods Impact of Black Rats on fruits and seeds At one or more sites (See Table 1) we established 5 plots beneath mature plants of several species (Howea forsteriana (Kentia Palm), Ochrosia elliptica (Berrywood), Olea paniculata (Maulwood), Coprosma putida) (Table 1). Each plot contained three treatments: uncaged; caged to exclude only birds (rodents could enter in gaps at each end (see Fig. 1); and caged to exclude both rodents and birds. At each plot, in each treatment we placed 5 mature fruits. The density of fruits so placed was not dissimilar to that found naturally below plants with mature fruits. We recorded the fate of seeds over 3-7 days. We also established an infared camera at one plot for each species to record what animal was responsible for seed removal or loss (these cameras were set up to activate and take 10 photos when a warm blooded animals comes into view of the camera (i.e. a rat , a mouse, a bird, a human). We also established plots for multiple species in a ‘cafeteria’ trial within the habitat where they generally occur. We did this for 12 species at 5 sites (See Table 1). In this trial 5 fruits or seeds of 4-6 species were placed together in small clumps (see Fig. 2). At each plot all seeds were uncaged. Again, like above, the fate of the fruits/seeds was followed over 3-7 days if possible. Impact of Black Rats on the Little Mountain Palm (Lepidorrhachis mooreana) on Mt Gower We repeated the sampling of Auld et al. (2010) (see Appendix 1) and sampled the size structure of stands of Little Mountain Palms in cloud forest on the Mount Gower plateau. We stratified the sampling between areas that have been baited by the Lord Howe Island Board to control rat numbers and those that have never been baited. In each of the baited and unbaited areas we sampled three separate plots for the Little Mountain Palm (Lepidorrhachis mooreana) (6 plots). We also sampled an additional 4 plots in the gradient between baited and unbaited areas. We estimated the distance for each of our 10 sampled plots from an existing bait station to determine if there was any additional benefit of baiting into surrounding unbaited areas. At each plot, we established a 5 m wide transect across the site and sampled all individual palms within the transect. The transects were sampled until we had encountered a minimum of 30-50 Lepidorrhachis individuals with an emergent trunk. We measured the trunk height to the base of the leaves, the number of leaf scars for individuals with a trunk and the height of individuals for individuals without a trunk. Extinction risk for the Gnarled Mossy Cloud Forest Ecological Community To assess the Gnarled Mossy Cloud Forest against the IUCN Red List for Ecosystems criteria (Keith et al. 2013), we searched the available literature for

4

evidence of any threats to the ecosystem that may result in ecosystem decline. We assessed the Gnarled Mossy Cloud Forest against four of the IUCN criteria (see Appendix 2, Auld and Leishman 2015 for more details):

• Decline was assessed using historical vegetation reports (Maiden 1898, 1914; Oliver 1916; Pickard 1983; Green 1994; Mueller-Dombois & Fosberg 1998; Harris et al. 2005). For more recent changes, we inspected satellite imagery (Google Earth and 2011 air photo imagery).

• For Restricted geographic distribution we used available Geographical Information System data layers for the distribution of Gnarled Mossy Cloud Forest (mapped from Pickard 1983) to estimate both the extent of occurrence and area of occupancy of the ecosystem.

• To assess changes in abiotic features that directly impact on Gnarled Mossy Cloud Forest, we used two metrics based on collated data from the Australian Bureau of Meteorology from mid 1940s onwards (Rainfall and cloud cover).

• For biotic degradation, we reviewed the impact of exotic rats on limiting recruitment of key species in the ecosystem and in changing the structure and composition of the ecosystem.

Results Impact of Black Rats on fruits and seeds In total, we were able to test the removal of fruits or seeds in 16 species, comprising a mixture of plant families, life forms (trees, shrubs, vines) and habitats. Most species examined had a fleshy fruit. In summary, seed or fruit losses were apparent in all study species, at least at some sites. Virtually all seed losses occurred at night and there was little removal of fleshy fruits by birds from our ground based trials. It is likely that most bird removal of fruits occurs on the plants and not the ground. Occasionally a shearwater returning to its nest disturbed one of our plots and scattered fruits, but this was not recorded as fruit loss in our calculations. Evidence of seed or fruit losses to rats was indicated by:

• Camera images of rats removing or consuming fruits/seeds (see Figs 3-6) • In situ seed or fruit remains indicating rats (eg, incisor marks etc) (see Figs 7-

11) We found no clear pattern of greater loss in areas without rat baiting stations present versus those with baiting stations. Trees Howea forsteriana There was high levels of seed removal in this species in all trials (5 sites, 2 time periods, Table 1), with 5/6 trials with 100% fruit loss and 1/6 with 80% loss. Olea paniculata All four trails across 4 different sites showed 100% seed loss. Syzygium fullagarii Fruit loss was around 40-44% at 2 study sites. For both sites, the trials only ran for 2-3 days so further fruit losses could be expected. Sarcomelicope simplicifolia

5

12% fruit loss at one site (insufficient fruits available for further study). Psychotria carronis 20% fruit loss at one site (trial only ran for 2 nights so further losses may occur). Chionanthus quadristamineus 32% fruit loss at one site (trial only ran for 2 nights so further losses may occur). Dysoxylon pachyphyllum 4% fruit loss at one site (trial only ran for 2 nights so further losses may occur). Drypetes deplanchei Seed loss ranged from 8-32% across 2 sites. Only fruits with no flesh available for trials. Sophora howinsula Seed loss ranged from 0-36% across four sites. When eaten seed, seeds were consumed in situ (see Fig. 8). Baloghia inophylla 100% seed loss in all trials across 4 sites. Shrubs Ochrosia elliptica Fruit loss varied from 4-96% across sites (Table 1). One of the sites with low losses was in a location with no adult plants. The cameras showed clear evidence of rats removing fruits (Fig. 4). On other occasions fruits were eaten in situ and fruit remains where present. Coprosma putida Fruit loss varied from 0-16% across 2 study sites. Dietes robinsoniana 36% fruit loss at one site (trial only ran for 2 nights so further losses may occur). Vines Jasminium simplicifolium Total fruit loss (100%) at three sites and 64% at the fourth Smilax australis Fruit loss varied from 92-100% across four sites. Geitonoplesium cymosum 68-100% fruit loss across 2 sites.

6

Impact of Black Rats on the Little Mountain Palm (Lepidorrhachis mooreana) on Mt Gower Based on sampling in 2006, Auld et al. (2010) showed that not only were the fruits of Lepidorrhachis mooreana consumed by rats, but the recruitment and establishment of a juvenile seedling bank only occurred in the baited habitats on Mt Gower. Our re-sampling of these and other sites in 2014 found the pattern remains unchanged. Two of the 3 sites that were unbaited lacked new seedlings and small juveniles (eg. Fig. 12) and one of these showed a decline in larger juveniles over time, suggesting ongoing recruitment failure. The unbaited site nearest the baiting stations showed a small pulse of juveniles less than 50 cm in height, suggesting that that there may have been some decline in rat impacts for a period in this area. The three sites in the baited area, all showed evidence of some small juveniles being present, but the abundance of small juveniles was greatly reduced in 2014 compared to 2006 (eg see Fig. 13). For the 4 additional sites sampled, one had no juvenile plants at all, two had no seedlings (one of these sites had only 1 juvenile less than 50 cm high), while only one site had a few seedlings. There was a pattern of greater recruitment of juvenile plants closer to bait stations (Fig. 14) with recruitment of small plants declining to zero at plots sampled around 250 m from bait stations. Extinction risk for the Gnarled Mossy Cloud Forest Ecological Community Overall the Gnarled Mossy Cloud Forest is considered to be Critically Endangered based on a restricted geographic distribution combined with continuing decline. Historical Decline: There have been no observed changes in the extent of this ecosystem in the last 50 years (Pickard 1983). No changes are indicated since Pickard’s 1983 vegetation classification based on inspection of recent satellite imagery (Google Earth and 2011 air photo imagery) and field traverses. The status of the ecosystem is therefore Least Concern under criterion A1. Geographical Extent and Continuing Decline: The geographic distribution of the ecosystem is below the IUCN threshold for Critically Endangered (Auld and Leishman 2015). Decline was inferred from: a loss of moisture from declining rainfall and cloud cover due to climate change (affecting disturbance regimes, gap formation and species survival and recruitment); ongoing exotic rat predation on seeds and seedlings of several sensitive species that are structural components of the ecosystem (affecting survival and recruitment); and the fact that the ecosystem is considered to exist at only one location (Auld and Leishman 2015). Decline in abiotic processes and features: There was a trend for decline in both rainfall and the frequency of cloud cover although there is still a great deal of variation in the degree of decline predicted (Auld and Leishman 2015, see Appendix 2). Decline in biotic processes and interactions: Rats damage key functional plant species in the ecosystem via predation on seeds and seedlings (Auld & Hutton 2004; Auld et al. 2010). Most of the area (93%) of Gnarled Mossy Cloud Forest is subject to impact by rats (7% is baited to reduce impact). The

7

extent and severity of seed predation by rats is high for two key species (mountain palms, but the severity of seed predation by rats on other taxa in the ecosystem is uncertain. More work on this aspect is required. Discussion Impact of Black Rats on fruits and seeds There was evidence that rats were consuming fruits or seeds in all 16 species examined. In summary, seed or fruit losses were apparent in all study species, at least at some sites. Losses were very high for six study species (Howea forsteriana, Olea paniculata, Baloghia inophylla, Jasminium simplicifolium, Smilax australis and Geitonoplesium cymosum); potentially very high but variable for one species (Ochrosia elliptica); moderate for three species (Syzygium fullagarii, Chionanthus quadristamineus, Dietes robinsoniana) (the actual losses may be higher as the trials only ran for a short period); generally low in 4 species (Sarcomelicope simplicifolia, Psychotria carronis, Dysoxylon pachyphyllum, Coprosma putida) (but the actual losses may be higher where the trials only ran for a short period); and low-moderate in two species (Sophora howinsula, Drypetes deplanchei). Further work on examining the impact of fruit losses on the ecology of the study species would assist interpretation of these data. Does the impact of these losses result in reduced potential for recruitment of new plants in the study species as has been previously demonstrated for the Little Mountain Palm (Lepidorrhachis mooreana) and Big Mountain Palm (Hedyscepe canterburyana) (Auld et al. 2010, Simmons et al. 2012)? Given that losses are occurring in all tested species, it would also be worthwhile to extend the sampling to additional species as many other species are also likely to be impacted by rats. Impact of Black Rats on the Little Mountain Palm (Lepidorrhachis mooreana) on Mt Gower Repeat sampling in 2014 confirmed the earlier findings of Auld et al. (2010) that Black Rats are significantly impacting on Little Mountain Palms on Mt Gower. Baiting appears to be effective in allowing the development of a juvenile bank of palms, although there was evidence that the size of this juvenile bank was greatly reduced in 2014 compared to 2006. Sites within about 100m of bait stations had some seedlings and small juveniles (indicating recent recruitment), but sites 250m or further away mostly had zero recent recruitment. This suggests that to successfully maintain this critically endangered palm on Mt Gower rat control or eradication across the Mt Gower summit will be required. Extinction risk for the Gnarled Mossy Cloud Forest Ecological Community Assessment of the Gnarled Mossy Cloud Forest against the IUCN Red List for Ecosystems criteria indicated that the ecosystem is critically endangered due to its very highly restricted geographic distribution (confined to summits of Mt Gower and Mt Lidgbird) combined with continuing decline due to a loss of moisture from declining rainfall and cloud cover as a result of climate change (affecting disturbance regimes, gap formation and species survival and recruitment); ongoing exotic rat predation on seeds and seedlings of several sensitive species that are structural components of the ecosystem (affecting survival and recruitment); and the fact that

8

the ecosystem is considered to exist at only one location (see Auld and Leishman 2015, Appendix 2 for details. This supports the current listing of this ecological community as critically endangered under the NSW Threatened Species Conservation Act (NSW Scientific Committee 2011). The recognition that the introduced Black Rat is a threat to the Gnarled Mossy Cloud Forest mirrors similar threats from exotic species and climate change to other Pacific island cloud forests (Meyer et al. 2010). Eradication of rats from Lord Howe Island will reduce the immediate risk to this ecosystem, however, only global mitigation of greenhouse gases could alleviate risk from declining cloud cover and moisture availability. If rat eradication does not occur or is unsuccessful, regular rodent baiting across much of the distribution of the ecosystem, particularly on Mt Gower would be needed to alleviate the threat. This may not be a long-term solution as the rodents may gain resistance to the poisons. A reduction in rat impact is needed to both allow successful seed production and recruitment of new plants in the Gnarled Mossy Cloud Forest. Acknowledgements We thank the Lord Howe Island Board for providing support to this project along with use of the Research Facility on the island (for TA, AD and MT). The assessment of extinction risk for the Gnarled Mossy Cloud Forest was supported by an Australian Research Council Linkage Grant to TA and others (LP LP130100435) and we thank Prof. David Keith and all the participants of the Australian Centre for Ecological Analysis and Synthesis workshop for their thoughtful contributions to the development of the risk assessment for this ecosystem. References Auld TD & Hutton I (2004) Conservation issues for the vascular flora of Lord Howe Island. Cunninghamia 8, 490–500. Auld TD, Hutton I, Ooi MKJ & Denham AJ (2010) Invasive species on oceanic islands: disruption of recruitment in narrow endemic palms. Biol. Invasions 12, 3351–61. Auld TD & Leishman MR (2015) Ecosystem risk assessment for Gnarled Mossy Cloud Forest, Lord Howe Island, Australia. Austral Ecology 40, 364–372. Billing J, Harden B (2000) Control of introduced Rattus rattus L. on Lord Howe Island. I. The response of mouse populations to warfarin bait used to control rats. Wildl Res 27, 655–658. Green PS (1994) Flora of Australia Vol. 49 Oceanic Islands 1. Australian Government Printing Service, Canberra. Harris R, Cassis G, Auld T & Hutton I (2005) Floristics and structure of the mossy cloud forest of Mt Gower summit, Lord Howe Island. Pac. Conserv. Biol. 11, 246–56. Hutton, I (1986) Lord Howe Island (Conservation Press: Canberra ACT).

9

Keith DA, Rodríguez JP, Rodríguez-Clark KM et al. (2013) Scientific Foundations for an IUCN Red List of Ecosystems. PLoS ONE 8, e62111. Lowry, PP (1998) Diversity, endemism, and extinction in the flora of New Caledonia: a review. In Rare, threatened and endangered floras of Asia and the Pacific Rim (eds. C.-I Peng & P.P. Lowry II), (Institute of Botany, Academia Sinicia Monograph Series No. 16. Taipei). pp. 181–206. Maiden JH (1898) Observations of the vegetation of Lord Howe Island. Proc. Linn. Soc. N. S.W. 23, 112–58. Maiden JH (1914) Further notes on the botany of Lord Howe Island (fifth paper). Proc. Linn. Soc. N. S.W. 39, 377–84. Meyer J-Y (2010) Montane cloud forests on remote islands of Oceania: the example of French Polynesia (South Pacific Ocean). In: Tropical Montane Cloud Forests: Science for Conservation and Management (eds L. A. Bruijnzeel, F. N. Scatena & L. S. Hamilton) pp. 121–9. Cambridge University Press, Cambridge. Mueller-Dombois D & Fosberg FR (1998) Vegetation of the Tropical Pacific Islands. Springer-Verlag, NewYork. NSW Scientific Committee (2010) Final determination. Solanum bauerianum - species presumed extinct listing [Cited 9th June 2015] Available from URL: http://www.environment.nsw.gov.au/determinations/solanumbauerianumFD.htm NSW Scientific Committee (2011) Final determination. Gnarled Mossy Cloud Forest on Lord Howe Island – critically endangered ecological community listing. [Cited 23 September 2014.] Available from URL: http://www.environment.nsw.gov.au/determinations/gnarledmossycloudforestlhifd.htm Oliver WRB (1916) The vegetation and flora of Lord Howe Island. Trans. Proc. N. Z. Inst. 49, 94–161. Pickard J (1983) Vegetation of Lord Howe Island. Cunninghamia 1, 133–265. Simmons L, Auld TD, Hutton I, Baker WJ, Shapcott A (2012) Will climate change, genetic and demographic variation or rat predation pose the greatest risk for persistence of an altitudinally distributed island endemic. Biology 1(3), 736-765. Wilkinson IS & Priddel D (2011) Rodent eradication on Lord Howe Island: challenges posed by people, livestock, and threatened endemics. In: Island Invasives: Eradication and Management (eds CR Veitch, MN Clout & DR Towns) pp. 508–14. IUCN, Gland.

10

Table 1 Seed and fruit losses across species and sites. Family Scientific

Name Common Name

Life form

Experiment Type

Site Date of trial

Seeds lost to rats (%)

Fleshy Fruit species Arecaceae Howea

forsteriana Kentia Palm

Palm Single Neds Beach

Nov-13

100

Howea forsteriana

Single Research Station

Nov-13

100

Howea forsteriana

Single Windy Ridge W

Nov-13

100

Howea forsteriana

Single Windy Ridge

Nov-13

100

Howea forsteriana

Single Windy Ridge E

Nov-13

80

Howea forsteriana

Single Neds Beach

Jan-14

100

Oleaceae Olea paniculata Maulwood Tree Single Research Station

Nov-13

100

Olea paniculata Single Middle Beach

Nov-13

100

Olea paniculata Single Windy Ridge W

Nov-13

100

Olea paniculata Single Windy Ridge E

Nov-13

100

Apocynaceae Ochrosia elliptica

Red Berrywood

Shrub Single Neds Beach

Nov-13

96

Ochrosia elliptica

Single Middle Beach

Nov-13

96

Ochrosia elliptica

Single Windy Ridge W

Nov-13

12

Ochrosia elliptica

Single Windy Ridge E

Nov-13

4

Myrtaceae Syzygium fullagarii

Scalybark Tree Single Research Station

Jun-14

44

Syzygium fullagarii

Cafeteria Erskine Valley

Jun-14

40

Rubiaceae Coprosma putida

Stinkwood Shrub Single Research Station

Jun-14

0

Coprosma putida

Single Erskine Valley

Jun-14

16

Oleaceae Jasminium simplicifolium

Jasmine Vine Cafeteria Research Station

Nov-13

64

Jasminium simplicifolium

Cafeteria Middle Beach

Nov-13

100

Jasminium simplicifolium

Cafeteria Windy Ridge E

Nov-13

100

Jasminium simplicifolium

Cafeteria Windy Ridge

Nov-13

100

Smilacaceae Smilax australis Vine Cafeteria Research Station

Nov-13

92

Smilax australis Cafeteria Middle Beach

Nov-13

100

Smilax australis Cafeteria Windy Ridge E

Nov-13

92

Smilax australis Cafeteria Windy Nov- 96

11

Ridge 13

Smilacaceae Geitonoplesium cymosum

Vine Cafeteria Research Station

Nov-13

68

Geitonoplesium cymosum

Cafeteria Middle Beach

Nov-13

100

Rutaceae Sarcomelicope simplicifolia

Yellow Wood

Tree Cafeteria Research Station

Nov-13

12

Rubiaceae Psychotria carronis

Black Grape

Tree Cafeteria Erskine Valley

Jun-14

20

Iridaceae Dietes robinsoniana

Wedding lily

Herb Cafeteria Erskine Valley

Jun-14

36

Oleaceae Chionanthus quadristamineus

Blue Plum Tree Cafeteria Erskine Valley

Jun-14

32

Meliacaea Dysoxylon pachyphyllum

Island Apple

Tree Cafeteria Erskine Valley

Jun-04

4

Old Fruit with no pulp Euphorbiaceae Drypetes

deplanchei Greybark Tree Cafeteria Research

Station Nov-13

8

Cafeteria Middle Beach

Nov-13

32

Dry seeds Fabaceae Sophora

howinsula Lignum Vitae

Tree Cafeteria Research Station

Nov-13

32

Sophora howinsula

Cafeteria Middle Beach

Nov-13

36

Sophora howinsula

Cafeteria Windy Ridge E

Nov-13

0

Sophora howinsula

Cafeteria Windy Ridge

Nov-13

0

Euphorbiaceae Baloghia

inophylla Bloodwood Tree Cafeteria Neds

beach Nov-13

100

Baloghia inophylla

Cafeteria Middle Beach

Nov-13

100

Baloghia inophylla

Cafeteria Windy Ridge E

Nov-13

100

Baloghia inophylla

Cafeteria Windy Ridge

Nov-13

100

12

Fig. 1 Layout of single treatment with fully enclosed cages, uncaged fruits and cage open at each end. Ochrosia elliptica fruits at Windy Pt west.

Fig. 2. Layout of cafeteria treatment. Species visible are (L to R): Smilax, Sarcomelicope, Baloghia.

13

Fig. 3. Black Rat removing mature fruit of Howea forsteriana near Ned’s Beach (Photo Ian Hutton)

Fig. 4. Black Rat removing fruit of Ochrosia elliptica from open ended cage.

14

Fig. 5. Black Rat removing fruit of Chionanthus quadristamineus fruit in Erskine Valley.

Fig 6. Black Rat fruit damage on Howea belmoreana (photo Ian Hutton)

15

Fig.7 Black Rat damage on Chionanthus quadristamineus in Erskine Valley

Fig. 8. Black Rat damage to Sophora howinsula seeds

16

Fig. 9 Black Rat damage to Olea paniculata fruits.

Fig. 10. Black Rat damage to Baloghia seeds.

Fig. 11. Black Rat damage to Smilax fruits and seeds

17

Fig. 12. A comparison of juvenile plants sizes between 2006 and 2014 at an unbaited site of Little Mountain Palm on Mt Gower. Note the lack of small juveniles and seedlings.

Fig. 13. A comparison of juvenile plants sizes between 2006 and 2014 at a baited site of Little Mountain Palm on Mt Gower. Note the presence of small juveniles and seedlings (particularly in 2006).

seedl

<50cm

<100cm

<150cm

<200cm

>200cm

Juvenile plant size (cm)

0

2

4

6

8

10

12

Freq

uenc

y

20142006

YEAR

seedl

<50cm

<100cm

<150cm

<200cm

>200cm

Juvenile plant size (cm)

0

100

200

300

400

500

Freq

uenc

y

20142006

YEAR

18

Fig. 14. Abundance of seedlings (black circles) and small juveniles (black triangles) of Little Mountain Palms in relation to distance from nearest bait station on Mt Gower.

0 100 200 300 400 500Distance to nearest bait station (m)

0

10

20

30

40

50

60

70

Abun

danc

e

0 100 200 300 400 500Distance to nearest bait station (m)

0

50

100

150

Abun

danc

e

19

Appendix 1 Auld TD, Hutton I, Ooi MKJ and Denham AJ (2010) Disruption of recruitment in two endemic palms on Lord Howe Island by invasive rats. Biological Invasions 12, 3351-61.

ORIGINAL PAPER

Disruption of recruitment in two endemic palmson Lord Howe Island by invasive rats

Tony D. Auld • Ian Hutton • Mark K. J. Ooi •

Andrew J. Denham

Received: 12 November 2009 / Accepted: 15 February 2010 / Published online: 6 March 2010

� Springer Science+Business Media B.V. 2010

Abstract Invasive species may have negative

impacts on many narrow range endemics and species

restricted to oceanic islands. Predicting recent

impacts of invasive species on long-lived trees is

difficult because the presence of adult plants may

mask population changes. We examined the impact

of introduced black rats (Rattus rattus) on two

palm species restricted to cloud forests and endemic

to Lord Howe Island, a small oceanic island in

the southern Pacific. We combined estimates of the

standing size distribution of these palms with the

proximal impacts of rats on fruit survival in areas

baited to control rats and in unbaited areas. The size

distribution of palms with trunks was comparable

across baited and unbaited sites. Small juvenile palms

lacking a trunk (\50 cm tall) were abundant in baited

areas, but rare in unbaited sites for Lepidorrhachis

mooreana, and rare or absent in 3 out of 4 unbaited

Hedyscepe canterburyana sites. All ripe fruits were

lost to rats in the small fruited L. mooreana. Fruit

removal was widespread but less (20–54%) in

H. canterburyana. Both palms showed evidence of

a reduced capacity to maintain a juvenile bank of

palms through regular recruitment as a consequence

of over 90 years of rat impact. This will limit the

ability of these species to take advantage of episodic

canopy gaps. Baiting for rat control reduced fruit

losses and resulted in the re-establishment of a

juvenile palm bank. Conservation of both endemic

palms necessitates control (or eradication) of rat

populations on the unique cloud forest summits of the

island.

Keywords Seed predation � Lord Howe Island �Australia � Palms � Black rat Rattus rattus

Introduction

Invasive species disrupt a range of ecological habitats

via direct impacts on individuals (either as adults or

juveniles) or by altering habitats, ecological pro-

cesses (Soule 1990; Lonsdale 1999) or disturbance

regimes (Mack and D’Antonio 1998). Invasive spe-

cies are generally competitors, predators, herbivores

or pathogens. Endemic populations of plants and

animals on oceanic islands often comprise few

individuals, occupy highly restricted areas or specia-

lised habitats and cannot recolonise from other

areas. These populations are particularly sensitive to

T. D. Auld (&) � A. J. Denham

Climate Change Science, Department of Environment,

Climate Change and Water (NSW), PO Box 1967,

Hurstville, NSW 2220, Australia

e-mail: tony.auld@environment.nsw.gov.au

I. Hutton

PO Box 157, Lord Howe Island, NSW 2898, Australia

M. K. J. Ooi

Department of Animal and Plant Sciences, University

of Sheffield, Sheffield S10 2TN, UK

123

Biol Invasions (2010) 12:3351–3361

DOI 10.1007/s10530-010-9728-5

declines or extinction caused by invasive species

(Sakai et al. 2002; Coote and Loeve 2003; Wiles et al.

2003; Blackburn et al. 2004; Towns et al. 2006;

Trevino et al. 2007), often in combination with loss

of habitat (Hunt 2007; Athens 2009).

In parts of the Pacific, much of the flora and fauna

have evolved in the absence of mammalian predators

and grazers. The subsequent introduction of such

predators, including humans, has led to a number of

serious declines or extinctions of species, including

birds (Hindwood 1940; Holdaway 1989; Blackburn

et al. 2004; Cheke 2006), invertebrates (Priddel et al.

2003; Coote and Loeve 2003; Towns et al. 2006) and

plants (Duncan and Young 2000; Campbell and

Atkinson 2002). For long-lived plants, extinctions

caused by introduced species have rarely been

recorded (Prebble and Dowe 2008), and in some

cases the life span of individual plants may be longer

than the period since the introduction of particular

weeds, predators or pathogens. For such plants,

continuing impacts may result in population declines,

even if these declines are hard to detect. Janzen

(1986) described such situations as blurry catastro-

phes and such species may be candidates for the

‘‘living dead’’ (sensu Janzen 1986) where a key part

of the life history has been disrupted and successful

reproduction has ceased. To identify recent impacts

of invasive species on long-lived plants we need to

determine if reproduction is sufficient to maintain a

stable population.

Past changes to population dynamics in plant

species have been inferred using paeleoenvironmen-

tal records (Hunt 2007; Athens 2009) or standing size

distributions across a range of plant communities

from arid shrublands (Crisp and Lange 1976; Walker

et al. 1986) to rainforests (Ogden 1985), including

palms (Enright 1985) and other species on oceanic

islands (Allen et al. 1994). To successfully apply

these techniques, an understanding of proximal

population dynamics is also needed. Assumptions

about growth rates or turnover in size or stage classes

must be made or estimated from direct observation

(Enright 1985; Ogden 1985). Proximate factors

limiting fecundity and survival, and the life or size

stages at which they act, need to be understood to

permit effective conservation management.

In this paper, we assess the impact of introduced

black rats (Rattus rattus) on two highly restricted

endemic palm species (Arecaceae) from Lord Howe

Island in the southern Pacific Ocean. Palms are a

useful study group as they are readily counted

and age classes may be inferred from leaf scars

(Tomlinson 1979). We inferred past impacts on

populations by examining standing size structures in

combination with some limited data on leaf produc-

tion. Concurrently, we estimated the proximal impact

of seed predation by rats across habitats either baited

(to reduce rat numbers) or unbaited. We applied

selective caging experiments to exclude vertebrates

from gaining access to seeds on the ground. Finally,

we combined these two approaches to infer the

conservation implication of the invasion of rats on the

endemic palm populations and the likely response of

these species to rat removal.

Methods

Study site

Lord Howe Island (31�300S, 159�050E) is a small

oceanic island (0.3–2.8 km wide and 11 km long,

1,455 ha) formed from volcanic activity 6.3–7 million

years ago (Hutton et al. 2007). It is situated *570 km

from the east coast of Australia, 1,350 km from New

Zealand and 1,250 km from New Caledonia (Pickard

1983a). Almost 80% of the island is protected in a

permanent park preserve (Davey 1986). It has been a

UNESCO World Heritage Area since 1982 (Depart-

ment of Environment and Climate Change NSW

2007). Topography on the island is dominated by its

two southern mountains (Mt Lidgbird at 777 m and

Mt Gower at 875 m). Both mountains are basaltic in

origin and Mt Gower is considered to be an eroded

caldera (McDougall et al. 1981).

The Lord Howe Island climate is humid-subtrop-

ical. It has a mean annual temperature of 19.2�C,

ranging from 17 to 25�C in December–February to

14–18�C in June–August (Mueller-Dumbois and

Fosberg 1998). Average annual rainfall in the low-

lands is 1,645 mm (Australian Bureau of Meterology),

although rainfall in the southern mountains is likely

to be much higher (Department of Environment and

Climate Change NSW 2007). Rainfall is unevenly

distributed throughout the year, with the driest period

in late summer (February, average of 100 mm)

and the wettest in winter (June/July average of

200 mm).

3352 T. D. Auld et al.

123

The island was originally free of mammals (except

for small insectivorous bats) when first encountered by

humans in 1788 and first settled in 1834 (Hutton 1986).

It now has a resident population of about 300 people

(Hutton et al. 2007). There is no archaeological

evidence of Polynesian occupation of the island prior

to 1788 (Anderson 2003) and the Pacific rat (Rattus

exulans) is not considered to have ever reached the

island. Black rats (Rattus rattus) were accidentally

introduced when a steamship ran aground in 1918

(Billing and Harden 2000). Rat control measures have

been sporadic until 1986 when a number of baited

areas were established and ongoing baiting was

maintained (Billing and Harden 2000). Approximately

8% of the island is currently baited (Billing 2000). This

rat control program is now part of an integrated

management plan for biodiversity conservation on

Lord Howe Island (Department of Environment and

Climate Change NSW 2007). The primary aim of the

baiting has been to reduce seed predation by rats upon

the two species of the endemic lowland palm genus

Howea, especially H. forsteriana which forms the

basis of a commercial seedling export enterprise on the

island for horticultural trade (Pickard 1980). The

baiting program also benefits other species, including

seabirds (McAllan et al. 2004), invertebrates (Depart-

ment of Environment and Climate Change NSW 2007)

and plants (Auld and Hutton 2004).

Study species and habitat

We sampled the two endemic mountain palm species

from Lord Howe Island. Lepidorrhachis mooreana

(F. Muell.) O.F. Cook (Arecoideae: Areceae;

Dransfield et al. 2008) is a small palm (trunks

generally up to 3 m) restricted to the cloud forest

vegetation (Pickard 1983a; Harris et al. 2005; Baker

and Hutton 2006) on the 27 Ha summit plateau of Mt

Gower (about 700–875 m in elevation). It has also

been reported from the very small summit of Mt

Lidgbird, but this requires confirmation. The species

is considered to be rare on Lord Howe Island (Pickard

1983b), with introduced rats and climate change

considered to threaten the cloud forest habitat of the

species (Baker and Hutton 2006). Hedyscepe canter-

buryana (C. Moore and F. Muell.) H Wend. & Drude

has a trunk of up to 6 m and is generally confined to

the higher elevations (600–875 m) of the southern

mountains, although on sheltered slopes it can be

found down to 300 m. It is listed as vulnerable under

the IUCN red list of threatened trees. It co-occurs

with L. mooreana on the summit of Mt Gower, but

also occurs at lower elevations in forests on several

extensive ledges below the summits of both Mt

Gower and Mt Lidgbird (Pickard 1983a). Since the

two palm species belong to monospecific genera we

hereafter refer to them as Hedyscepe and

Lepidorrhachis.

The size and dry weight (air dried at room

temperature) of fruits of the species are very differ-

ent. Lepidorrhachis fruits are globular drupes

*1.3 cm in diameter, weighing an average of

1.4 ± 0.04 g (n = 20 fruits). Fruits of Hedyscepe

are large and ovoid, *3 9 4.4 cm, weighing 24.8

± 0.5 g (n = 30 fruits).

Standing size structure

We sampled the size structure of stands of the two

species in cloud forest on the Mount Gower plateau.

We stratified the sampling between areas that have

been baited by the Lord Howe Island Board to control

rat numbers and those that have never been baited.

Baiting currently involves placement of poison every

3 months at fixed stations throughout the baited area.

Baiting has been conducted on Mt Gower for the last

20 years (Billing 2000). Only a small proportion

(*7%) of the eastern area of the summit has been

baited (1.9 ha of a total 27 ha) and not all the baited

or unbaited habitat is occupied by the study species.

Although both mountain palms occur across the

summit plateau and Hedyscepe and Lepidorrhachis

co-occur in places, there is a tendency for Hedyscepe

to occupy more sheltered sites and gully slopes while

Lepidorrhachis is more common on small ridge tops.

In each of the baited and unbaited areas we sampled

three separate plots for each palm species (12 plots

overall). At each plot, we established a 5 m wide

transect across the site and sampled all individual

palms within the transect. The transects were sampled

until we had encountered a minimum of 30 Hedy-

scepe or 50 Lepidorrhachis individuals with an

emergent trunk. Hence, the number of juvenile plants

without a trunk that were sampled varied between

plots. Transects ran for a minimum of 20 m up to

35 m, but occasionally to obtain sufficient numbers

of plants with trunks we had to sample additional

trunked plants beyond this length. These latter plants

Disruption of recruitment in two endemic palms on Lord Howe Island 3353

123

were not included in our estimates of plant densities.

We measured canopy height (=leaf height in plants

lacking a trunk) for all individuals. For plants with

emergent trunks, we also scored flowering or fruiting

and measured two additional plant size attributes:

1. trunk height to the base of the leaf sheaths; and

2. the number of leaf scars on the trunk.

Occasionally, trunks were covered in mosses and

lichens and it was not possible to count the number of

leaf scars on the entire trunk. Where this occurred,

mainly in Hedyscepe, we counted a section of the

trunk and extrapolated to the length of the trunk. This

assumed even rates of growth between leaf scars. In

Hedyscepe and Lepidorrhachis, there was a pattern of

larger intervals between leaf scars at the base of

trunks and smaller intervals at the top (after the plants

had matured). Consequently, where we had to

subsample a trunk, we chose a section that included

the transition between smaller and larger intervals

between leaf scars. We also calculated the average

leaf scar interval per trunked plant from the stem

length divided by number of leaf scars.

We additionally sampled one population of Hedy-

scepe at the summit of Mt Lidgbird. The summit of

this mountain is very narrow and there is little

available habitat. The habitat of Hedyscepe palms on

this mountain top has not been subject to rat baiting.

Sufficient habitat was available to sample one plot,

where we ran a 5 m transect along the undulating

summit and sampled as per above. It was not possible

to sample Lepidorrhachis on Mt Lidgbird as no plants

could be located.

Hedyscepe also occurs on large ledges below the

summits of Mt Lidgbird and Mt Gower. However, to

ensure that all our samples were confined to compa-

rable habitat on the mountain summits in cloud forest

vegetation, we did not sample in these areas.

Inferring plant age or growth rates

For Lepidorrhachis, we tagged leaf bases on 10

mature trunked plants to determine the rate of

production of leaf scars over a 31 month period.

The length of time a palm had a trunk was then

inferred from the mean number of leaves produced

per year times the number of leaf scars. All plants

sampled were on ridges on the summit of Mt Gower

across the gradient in elevation. Loss of tags from

Hedyscepe plants prevented a similar estimation in

that species.

Impact of rats on fruit survival

To examine the direct proximal impacts of rats on the

palms we established seed predation exclusion exper-

iments on Mt Gower. We ran the experiments on

2 occasions in 2006 (winter–July; and late spring–

November). This replication allowed us to examine

temporal variation in seed predation rates across the

long period of fruit maturation. For both replicate

trials, there were developing and ripe fruits available

on mature plants of both species. For Lepidorrhachis,

there were considerably more ripe fruits in July than

November, while for Hedyscepe, slightly more ripe

fruits were available in November. At each time, we

sampled 4 sites per species, 2 in rat baited and 2 in

unbaited areas. At each site, we established 10

randomised plots of treatments, with the location of

each plot constrained by being under a palm with a

fruiting infructescence, but such plants were chosen

randomly from those available at the site. For Hedy-

scepe, at each chosen palm, we set up 2 batches of 10

mature fruits in a cache, one caged and one uncaged.

The caged fruits were covered with a 1 cm2 wire mesh

cage that was held into place by wire stakes. This

excluded rodents and birds. We followed the fates of

fruits over a 2 week period. For Lepidorrhachis, we

observed that green fruits were eaten by rats. In some

baited areas, fruits, in which the red fleshy outer layer

(the mesocarp and exocarp) had been removed or worn

off, persisted on the forest floor (leaving only the seed

and endocarp). So for Lepidorrhachis, we used caged

and uncaged ripe red fruits along with two additional

uncaged fruit categories at each plot. These categories

were large green fruit and endocarp-only fruit. We

observed Lord Howe currawongs (Strepera graculina

crissalis, a diurnal corvid) remove ripe fruits from

infructescences of Lepidorrhachis and regurgitate

endocarps. As this bird may have removed fruits in

our exclusion trials, the inclusion of endocarp-only

fruits in the experiment allowed us to compare

removal rates for fruits that were attractive to these

birds (ripe red fruits) and fruits that were not (endo-

carp-only). The Lord Howe currawongs may occa-

sionally handle the much larger fruits of Hedyscepe

(Hutton pers observ.), but no endocarp-only fruits

were available on the summit plateau.

3354 T. D. Auld et al.

123

In the July trial, the experiments were checked

overnight and again at *2 weeks after establishment,

with the final data point used in the analyses. In the

November trial, we checked removal and fruit loss at

plots overnight, at 3–3.5 days and after 2 weeks. We

used these temporal sampling data to examine the early

rates of fruit loss.

To infer the fates of fruits we explored several

possibilities. We spent time observing fruit caches in

the day and night to monitor removal agents. We also

examined the pattern of fruit mesocarp removal and

fruit consumption in situ to infer likely agents.

Data analysis

We compared the population size structure in stands of

each palm across baited and unbaited areas. For each

species, we used nested 2 Factor ANOVAs (fixed

factor rat baiting, with sites nested in rat baiting) to

compare each size attribute in trunked plants. Species

were not included as a factor as Hedyscepe is much

larger than Lepidorrhachis and as we were interested in

the nested site effect within a species. Cochran’s Test

was used to test for homogeneity of variances. Where

heterogeneous variances were detected, the data were

log transformed (Underwood 1997). Where ANOVAs

were significant, individual means were compared

using Student–Newman–Keuls tests (Zar 1974).

For the predator exclusion experiment, for each

species, we compared the magnitude of fruit removal/

loss across the treatment plots. We used a GLIM with

four factors—time of experiment, rat baiting, caging

treatment and sites (nested) with a binomial error

structure. Removal from a plot was scored if there

had been any seed removal as the plot was the unit of

replication. In the November trial, we used a failure-

time analysis (Fox 2001) to compare the rates of seed

removal between baited and unbaited areas and

different caging treatments, pooling data across plots

and sites.

Results

Standing size structure

In trunked palms of both Hedyscepe and Lepidorrhachis,

there was no significant difference between baited and

unbaited areas for all size attributes, but there was

variation among study sites (Table 1). In Hedyscepe,

there was a significant difference in crown sizes

(F4,135 = 3.3, P = 0.01), trunk lengths (F4,177 = 5.4,

P \ 0.001) and the number of leaf scars per trunk

(F4,170 = 2.7, P = 0.03) across sites (Table 1). In

Lepidorrhachis, there was significant variation among

sites in crown sizes (F4,211 = 3.5, P = 0.001) and

trunk lengths (F4,294 = 2.8, P = 0.025).

The size distribution of trunked plants was broadly

similar across sites, although in Hedyscepe there were

relatively more individuals in the largest size classes

than in Lepidorrhachis (Figs. 1, 2). When plants

without trunks were compared, there were clear

differences between the population structures in the

rat baited versus unbaited areas (Figs. 1, 2). Small

juvenile palms (\25 cm) were only present in areas

that were baited, with the exception of one Hedyscepe

site (H2rat; Fig. 1). Large juveniles that did not yet

have an aerial trunk and plants with a trunk that had

clearly not yet flowered were present at all sites,

although they were very scarce at one Hedyscepe site

where rats were not baited (H1rat, Table 2). The

Hedyscepe site from the summit of Mt Lidgbird was

comparable to the unbaited Mt Gower sites as there

were few plants without a trunk (Table 2). Densities

of mature (0.06–0.4 m-2) and juvenile plants (0.05–

8.3 m-2) varied across sites (Table 2).

Trunked Lepidorrhachis plants produced an aver-

age of 2.6 (±0.8) leaves each year (range 2.3–3.1).

To infer an estimate of the time spent as a trunked

Lepidorrhachis palm, we assumed plants produced

between 2 and 3 leaf scars per year. This would

suggest that each site contains a few very old

individuals that have had trunks for 100–200 years

(cf. Fig. 2). Most of the plants are estimated to have

had trunks for less than 70 years.

Impact of rats on fruit survival

Fruit loss varied markedly between baited and

unbaited areas. For Hedyscepe, there was no signif-

icant difference between the replicate experiment

times and no significant interactions involving this

term. No fruits were consumed in situ in this species.

There were significant differences between baited and

unbaited areas and between caging treatments, along

with significant site variation. Some fruits were

removed from all plots at one of the unbaited sites

Disruption of recruitment in two endemic palms on Lord Howe Island 3355

123

(a total 54% of individual fruits removed) and from

85% of plots at the other unbaited site (a total 20% of

individual fruits removed). One of the baited sites

also had some fruit removed from 85% of plots

(a total 22% of individual fruits removed), while the

second baited site had removal from 45% of plots

(a total 7% of individual fruits removed). There was

evidence of rats scraping away the thin mesocarp and

trying to remove the fibrous endocarp. Nocturnal

observations revealed that rats dragged fruits away

from the experimental plots. It is likely that fruits

were consumed in caches and such caches with eaten

fruits were observed in the habitat.

In Lepidorrhachis, there was a significant two way

interaction between caging treatment and date of

sampling (July or November), along with a signifi-

cant difference between baited and unbaited areas

and a significant nested site effect. In the unbaited

areas, uncaged fruits were removed in all plots,

except for 2 plots of green fruits. All red fruits and

those with the flesh removed (endocarp-only) were

taken in both July and November sampling. The

removal of green fruits varied from 98% in July to

17% in November. Evidence that rats consumed

fruits was indicated by the presence of chewed pieces

of pericarps in situ, the presence of fresh rat scats in

the plots and our nocturnal observations of rats

feeding on fruits. While some ripe red fruits may

have been taken and dispersed by Lord Howe

currawongs, it is likely that rats are consuming fruits

since there was a loss of both ripe red fruits and

endocarp-only fruits in all the unbaited sites. Fruits

with the mesocarp removed should not be attractive

to currawongs. At these unbaited sites, no adult plants

had ripe fruits on infructescences and no fruits with

the mesocarp removed (endocarp-only) were present

on the ground. In contrast, ripe fruits were common

both on infructescences and on the ground in the

baited areas. In baited areas, we found that there was

some removal of ripe red fruits in all plots in

November (a total 64–94% of individual fruits

removed across sites) and in 85% of plots in July

(a total 35–40% of individual fruits removed across

sites). For fruits with only an endocarp, there was

more variation between the two sampling seasons,

with removal in 95% of plots in November (a total

24–96% of individual fruits removed across sites)

compared to only 30% of plots in July (a total 6% of

individual fruits removed across sites). Removal of

green fruits was comparable across sampling seasons,

65% of plots in November (a total 14–25% of

individual fruits removed across sites) and 55% of

plots in July (a total 12–13% of individual fruits

removed across sites). The losses in November at one

baited site were directly attributed to rats as evi-

denced by large amounts of fruit remains in situ. This

Table 1 Summary size measurements for trunked palms sampled on Lord Howe Island

Species Site Baited for rats Trunk Length (m) Mean plant (±SE)

Crown size (m) Leaf scars Leaf scar

interval (cm)

Hedyscepe H1rat No 4.01(0.34) 2.08(0.19) 193(26) 2.7(0.3)

H2rat No 2.73(0.33) 1.98(0.10) 138(32) 3.4(0.3)

H3rat No 2.05(0.27) 2.38(0.10) 87(22) 4.2(0.5)

H4 Yes 2.63(0.33) 2.43(0.10) 115(25) 4.1(0.4)

H5 Yes 2.39(0.35) 2.38(0.13) 148(34) 4.5(0.5)

H6 Yes 3.03(0.32) 2.03(0.09) 208(39) 3.1(0.4)

HLrat No 2.79(0.21) 1.89(0.07) 170(14) 2.3(0.2)

Lepidorrhachis L1rat No 1.37(0.14) 1.73(0.05) 75(9) 2.1(0.2)

L2rat No 1.67(0.15) 1.85(0.08) 68(9) 2.5(0.2)

L3rat No 1.59(0.12) 1.63(0.07) 70(8) 2.4(0.1)

L4 Yes 1.73(0.12) 2.02(0.06) 76(9) 2.8(0.2)

L5 Yes 1.21(0.13) 1.95(0.08) 53(9) 2.9(0.3)

L6 Yes 1.46(0.11) 1.80(0.05) 75(9) 2.3(0.1)

For each size measurement 30 Hedyscpe or 50 Lepidorrhachis trunked plants were sampled

3356 T. D. Auld et al.

123

may reflect re-colonisation of this site by rats at the

time.

In the November trial, there was some initial

(overnight) rapid removal of ripe fruits of Hedyscepe

in unbaited areas and then a slow rate of removal in

both baited and unbaited areas (Fig. 3). In Lep-

idorrhachis, loss of ripe red fruits was rapid where

rats were not baited (Fig. 4), with some 90% fruits

lost within a day and all fruits gone within 3 days.

Most of these fruits were consumed in situ by rats

over the first night of the experiment, again support-

ing the conclusion that consumption by rats rather

than fruit removal and dispersal by Lord Howe

currawongs explains fruit losses. The loss of all

endocarp-only fruits within 3 days was also likely

due to rat predation since these fruits are not

attractive to currawongs (Fig. 4).

Discussion

The introduction of rats (Rattus rattus) to the oceanic

Lord Howe Island is likely to have increased the risk

of extinction for the two endemic mountain palms.

This is a consequence of rat predation of fruits which

has the potential to limit recruitment in both palm

species. Past observations highlight the lack of ripe

fruits on Lepidorrhachis plants unless mesh caging

was applied to exclude rats from developing fruits

(Moore 1966; Pickard 1980). The paucity of small

0 100 200 3000

10

20

30

0 100 200 300 400 500 600 7000

10

20

30Fr

eque

ncy

Leaf Height (cm) Number of Leaf scars

0 100 200 3000

100

200

300

0 100 200 300 400 500 600 7000

10

20

30

Freq

uenc

y

Leaf Height (cm) Number of Leaf scars

0 100 200 3000

100

200

300

0 100 200 300 400 500 600 7000

10

20

30

Freq

uenc

y

Leaf Height (cm) Number of Leaf scars0 100 200 300

0

100

200

300

0 100 200 300 400 500 600 7000

10

20

30

Freq

uenc

yLeaf Height (cm) Number of Leaf scars

0 100 200 3000

10

20

30

0 100 200 300 400 500 600 7000

10

20

30

Freq

uenc

y

Leaf Height (cm) Number of Leaf scars0 100 200 300

0

100

200

300

0 100 200 300 400 500 600 7000

10

20

30

Freq

uenc

y

Leaf Height (cm) Number of Leaf scars

H1rat

H2rat

H3rat

H4

H5

H6

Fig. 1 Size distributions for Hedyscepe canterburyana of plant height for juvenile plants lacking a trunk (open bars) and the number

of leaf scars in trunked individuals (filled bars). rat No rat baiting conducted at the site. Note: Y axis scale varies across sites

Disruption of recruitment in two endemic palms on Lord Howe Island 3357

123

juvenile plants in areas which have not been baited

for rat control, both in our study plots and more

generally throughout this unbaited area, supports the

more general contention that the introduction of rats

affects plant recruitment in a range of island ecosys-

tems (Campbell and Atkinson 2002; Delgado Garcıa

2002; Towns et al. 2006; Meyer and Butaud 2009;

Traveset et al. 2009). It is possible that rats also

reduce seedling and juvenile growth as they have

been observed eating palm leaf sheaths and causing

leaf fall in both juvenile and adult palms consistent

with similar damage observed on plants elsewhere

(Traveset et al. 2009). The impact of rats is greatest in

Lepidorrhachis, where fruit losses reached 100% and

small juvenile plants (\50 cm) were extremely rare

in the presence of rats. Any possible seed escape in

Lepidorrhachis via dispersal by the Lord Howe

currawong is negated by the loss of endocarp-only

fruits (matching regurgitated fruits) in unbaited areas.

For Hedyscepe, observed fruit losses were less (total

of 20–54%), and while some seed escape may be

occurring, only one out of four sites sampled on the

mountain tops has an appreciable number of small

juveniles (\50 cm; Table 2). Differences between

the two palm species may relate to the fact that the

fruits of Hedyscepe are up to 18 times bigger than

Lepidorrhachis and rats may have more difficulty

eating into Hedyscepe fruits as they have a thicker

and very fibrous endocarp wall. Neither palm species

can be considered the ‘‘living dead’’ (sensu Janzen

0 100 200 3000

10

20

30

0 100 200 300 400 5000

10

20

30Fr

eque

ncy

Leaf Height (cm) Number of Leaf scars0 100 200 300

0

100

200

300

400

500

600

700

0 100 200 300 400 5000

10

20

30

Freq

uenc

y

Leaf Height (cm) Number of Leaf scars

0 100 200 3000

10

20

30

0 100 200 300 400 5000

10

20

30

Freq

uenc

y

Leaf Height (cm) Number of Leaf scars0 100 200 300

0

100

200

300

400

0 100 200 300 400 5000

10

20

30

Freq

uenc

y

Leaf Height (cm) Number of Leaf scars

0 100 200 3000

10

20

30

0 100 200 300 400 5000

10

20

30

Freq

uenc

y

Leaf Height (cm) Number of Leaf scars0 100 200 300

0

100

200

300

400

0 100 200 300 400 5000

10

20

30

Freq

uenc

y

Leaf Height (cm) Number of Leaf scars

L1rat

L2rat

L3rat

L4

L5

L6

Fig. 2 Size distributions for Lepidorrhachis mooreana of plant height for juvenile plants lacking a trunk (open bars) and the number

of leaf scars in trunked individuals (filled bars). rat No rat baiting conducted at the site. Note: Y axis scale varies across sites

3358 T. D. Auld et al.

123

1986) as there are juvenile plants over 50 cm in

height at all sites, independent of the current rat

baiting program on the island. Some episodic

recruitment may occur as a result of sporadic seed

escape from rats, possibly in relation to natural

fluctuations of abundance within the rat population.

Rats reached Lord Howe Island some 90 years

prior to this study and are reported to have spread

rapidly (Billing and Harden 2000). Consequently, rat

predation could have affected palm recruitment for a

number of decades. Is this reflected in the estimated

Table 2 Plant densities across sampled plots for Lepidorrhachis mooreana and Hedyscepe canterburyana

Species Site Baited for rats Density (m-2)

Trunked plants Juveniles with no trunks

Mature Juv Height (m)

\0.5 0.5–1 [1

Hedyscepe H1rat No 0.10 0.02 0 0.01 0.01

H2rat No 0.13 0.02 0.85 0.01 0.03

H3rat No 0.11 0.14 0.03 0.03 0.18

H4 Yes 0.14 0.07 4.02 0.56 0.19

H5 Yes 0.06 0.09 1.52 0.07 0.06

H6 Yes 0.08 0.03 2.91 0.11 0.03

HLrat No 0.40 0.11 0.01 0.02 0.04

Lepidorrhachis L1rat No 0.34 0.06 0.04 0.09 0.10

L2rat No 0.24 0.06 0.01 0.09 0.15

L3rat No 0.35 0.04 0.01 0.04 0.11

L4 Yes 0.19 0 8.13 0.13 0.08

L5 Yes 0.12 0.10 2.38 0.19 0.11

L6 Yes 0.25 0.04 5.41 0.28 0.14

All sites on Mt Gower, except HLrat from Mt Lidgbird

0 5 10 15 20DAY

0.0

0.1

0.2

0.3

0.4

0.5

Frui

t rem

oval

(pr

op.)

Fig. 3 Rate of fruit loss in Hedyscepe canterburyana, pooled

across sites. Data are 1-(Kaplan–Meier estimates) with

standard errors. Filled symbols are sites with no rat baiting,

symbols are sites with rat baiting. Vertical bars are standard

errors. Treatments are: Cage (circle); uncaged ripe fruit

(triangle). No fruits were removed in the caged treatment

0 5 10 15 20DAY

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Frui

t rem

oval

(pr

op.)

Fig. 4 Rate of fruit loss in Lepidorrhachis mooreana, pooled

across sites. Data are 1-(Kaplan–Meier estimates) with

standard errors. Filled symbols are sites with no rat baiting,

open symbols are sites with rat baiting. Vertical bars are

standard errors. Treatments are: Cage (circle); uncaged ripe

fruit (triangle); uncaged green fruit (diamond) uncaged

endocarp-only fruit (square)

Disruption of recruitment in two endemic palms on Lord Howe Island 3359

123

age structure of Lepidorrhachis? Plant age in palms

has been inferred by the number of leaf scars

(Sarukhan 1980; Enright and Watson 1992). In

Lepidorrhachis, there were few plants (*6% of the

population) with more than 120 leaf scars (Fig. 2).

This represents some 40–60 years of trunk growth.

The remainder of the trunked palms are more or less

evenly distributed up to 100 or 120 leaf scars (Fig. 2).

If individual mortality was constant across these size

classes then this would suggest a declining population

and may reflect a long term impact of rats on

recruitment. Estimating plant age requires additional

data on the time required for juvenile plants to grow

large enough to produce a trunk, and this may vary

with habitat quality and disturbance regime. This pre-

trunk period typically takes decades in some palm

species (e.g. 35–60 years, Enright 1985; Enright and

Watson 1992), but is currently unknown for Lep-

idorrhachis. Further data on juvenile survival and

growth are needed to clarify the age structure of the

population.

Recovery of vegetation following removal of rats

is not always straightforward (Mulder et al. 2009;

Towns 2009). Rats may impact on nutrient cycling

(through impacts on seabirds, Mulder et al. 2009) and

the impacts of mice may increase if rats alone are

removed (Angel et al. 2009). In our study area on Mt

Gower, there has only been limited rat control in the

last 20 years. There is a lack of small juveniles in

unbaited areas at all sampled Lepidorrhachis sites

and at 75% of the Hedyscepe sites. Baited areas are

distinguished by the abundance of young juvenile

palms, implying a restoration of plant recruitment

and a re-establishment of the juvenile bank of

palms (sensu Silvertown 1982). Successful growth

to maturity of juvenile palms will probably be

influenced by local canopy gap formation after tree

death or tree fall in storms.

Long-term conservation of the endemic palm

genera in this study may be compromised by a

reduced capacity to establish a juvenile bank and to

adapt to a changing climate. Both species occur in the

unique cloud forests on the island (Harris et al. 2005)

and changes to moisture regimes (e.g. through a

reduction in cloud cover) with a changing climate

may limit establishment opportunities in the future.

Targeted control, or eradication of rats is essential to

maximise the potential for long-term conservation of

these species.

Acknowledgments We thank the Lord Howe Island Board

for permission to carry out this study and for support in the use

of a research facility for TA, MO and AD. William Baker made

helpful comments on the manuscript.

References

Allen RB, Lee WG, Rance BD (1994) Regeneration in indig-

enous forest after eradication of Norway rats, Breaksea

Island, New Zealand. NZ J Bot 32:429–439

Anderson A (2003) Investigating early settlement on Lord

Howe Island. Aust Archaeol 57:98–102

Angel A, Wanless RM, Cooper J (2009) Review of impacts of the

introduced house mouse on islands in the Southern Ocean:

are mice equivalent to rats? Biol Invas 11:1743–1754

Athens JS (2009) Rattus exulans and the catastrophic disap-

pearance of Hawai’i’s native lowland forest. Biol Invas

11:1489–1501

Auld TD, Hutton I (2004) Conservation issues for the vascular

flora of Lord Howe Island. Cunninghamia 8:490–500

Baker WJ, Hutton I (2006) Lepidorrhachis. Palms 50:33–38

Billing J (2000) The control of introduced Rattus rattus L. on

Lord Howe Island. II. The status of warfarin resistance in

rats and mice. Wildl Res 27:659–661

Billing J, Harden B (2000) Control of introduced Rattus rattusL. on Lord Howe Island. I. The response of mouse pop-

ulations to warfarin bait used to control rats. Wildl Res

27:655–658

Blackburn TM, Cassey P, Duncan RP, Evans KL, Gaston KJ

(2004) Avian extinction and mammalian introductions on

oceanic islands. Science 305:1955–1958

Campbell DJ, Atkinson IAE (2002) Depression of tree

recruitment by the Pacific rat (Rattus exulans Peale) on

New Zealand’s northern offshore islands. Biol Conserv

107:19–35

Cheke AS (2006) Establishing extinction dates—the curious

case of the Dodo Raphus cucullatus and the Red Hen

Aphanapteryx bonasia. Ibis 148:155–158

Coote T, Loeve E (2003) From 61 species to five: endemic tree

snails of the Society Islands fall prey to an ill-judged

biological control programme. Oryx 37:91–96

Crisp MD, Lange RT (1976) Age structure, distribution and

survival under grazing of the arid-zone shrub Acaciaburkittii. Oikos 27:86–92

Davey A (1986) Plan of management. Lord Howe Island per-

manent park preserve. NSW National Parks and Wildlife

Service, Sydney

Delgado Garcıa JD (2002) Interaction between introduced rats

and a frugivore bird–plant system in a relict island forest.

J Nat Hist 36:1247–1258

Department of Environment and Climate Change NSW (2007)

Lord Howe Island Biodiversity Management Plan.

Department of Environment and Climate Change (NSW),

Sydney

Dransfield J, Uhl NW, Asmussen CB, Baker WJ, Harley MM,

Lewis CE (2008) Genera Palmarum. The evolution and

classification of palms. Kew Publishing, UK

Duncan RP, Young JR (2000) Determinants of plant extinction

and rarity 145 years after European settlement of Auck-

land, New Zealand. Ecology 81:3048–3061

3360 T. D. Auld et al.

123

Enright NJ (1985) Age, reproduction and biomass allocation in

Rhopalostylis sapida (Nikau palm). Aust J Ecol 10:461–

467

Enright NJ, Watson AD (1992) Population dynamics of the

Nikau palm, Rhopalostylis sapida (Wendl et Drude), in a

temperate forest remnant near Auckland, New Zealand.

NZ J Bot 30:29–43

Fox GA (2001) Failure-time analysis: Studying times to events

and rates at which events occur. In: Scheiner SM, Gu-

revitch J (eds) Design and analysis of ecological experi-

ments. Oxford University Press, Oxford, pp 235–266

Harris R, Cassis G, Auld TD, Hutton I (2005) Patterns in flo-

ristic composition and structure in the mossy cloud forest

of Mt Gower summit, Lord Howe Island. Pac Conserv

Biol 11:246–256

Hindwood KA (1940) The birds of Lord Howe Island. Emu

40:1–86

Holdaway RN (1989) New Zealand’s pre-human avifauna and

its vulnerability. NZ J Ecol 12:11–25

Hunt TL (2007) Rethinking Easter Island’s ecological catas-

trophe. J Archaeol Sci 34:485–502

Hutton I (1986) Lord Howe Island. Conservation Press, Can-

berra, ACT

Hutton I, Parkes JP, Sinclair ARE (2007) Reassembling island

ecosystems: the case of Lord Howe Island. Anim Cons

10:22–29

Janzen DH (1986) Blurry catastrophes. Oikos 47:1–2

Lonsdale WM (1999) Global patterns of plant invasions and

the concept of invasibility. Ecology 80:1522–1536

Mack MC, D’Antonio CM (1998) Impacts of biological inva-

sions on disturbance regimes. Trends Ecol Evol 13:195–

198

McAllan IAW, Curtis BR, Hutton I, Cooper RM (2004) The

birds of Lord Howe Island Group: a review of records.

Aust Field Ornithologist 21(Suppl):1–82

McDougall I, Embleton BJ, Stone DB (1981) Origin and

evolution of Lord Howe Island, southwest Pacific Ocean.

J Geol Soc Aust 28:155–176

Meyer JY, Butaud JF (2009) The impacts of rats on the

endangered native flora of French Polynesia (Pacific

Islands): drivers of plant extinction or coup de grace

species? Biol Invas 11:1569–1585

Moore HE (1966) Palm hunting around the world. IV Lord

Howe Island. Principes 10:13–21

Mueller-Dumbois D, Fosberg FR (1998) Vegetation of the

tropical Pacific islands. Springer, New York

Mulder CPH, Grant-Hoffman MN, Towns DR, Bellingham PJ,

Wardle DA, Durrett MS, Fukami T, Bonner KI (2009)

Direct and indirect effects of rats: does rat eradication

restore ecosystem functioning of New Zealand seabird

islands? Biol Invas 11:1671–1688

Ogden J (1985) An introduction to plant demography with

special reference to New Zealand trees. NZ J Bot 23:751–

772

Pickard J (1980) The palm seed industry of Lord Howe Island.

Principes 24:3–13

Pickard J (1983a) Vegetation of Lord Howe Island. Cun-

ninghamia 1:133–266

Pickard J (1983b) Rare or threatened vascular plants of Lord

Howe Island. Biol Conserv 27:125–139

Prebble M, Dowe JL (2008) The late Quaternary decline and

extinction of palms on oceanic Pacific islands. Quatern

Sci Rev 27:2546–2567

Priddel D, Carlile N, Humphrey M, Fellenberg S, Hiscox D

(2003) Rediscovery of the ‘extinct’ Lord Howe Island

stick-insect (Dryococelus australis (Montrouzier)) (Phas-

matodea) and recommendations for its conservation.

Biodiversity Conserv 12:1391–1403

Sakai AK, Wagner WL, Mehrhoff LA (2002) Patterns of

endangerment in the Hawaiian flora. Syst Biol 51:276–

302

Sarukhan J (1980) Demographic problems in tropical systems.

In: Solbrig O (ed) Demography and evolution in plant

populations. Blackwell Scientific Publications, Oxford, pp

161–188

Silvertown JW (1982) Introduction to plant population ecol-

ogy. Longman Press, London

Soule ME (1990) The onslaught of alien species, and other

challenges in the coming decades. Conserv Biol 4:233–

239

Tomlinson PB (1979) Systematics and ecology of the Palmae.

Annu Rev Ecol Syst 10:85–107

Towns DR (2009) Eradications as reverse invasions: lessons

from Pacific rat (Rattus exulans) removals on New Zealand

islands. Biol Invas 11:1719–1733

Towns DR, Atkinson IAE, Daugherty CH (2006) Have the

harmful effects of introduced rats on islands been exag-

gerated? Biol Invas 8:863–891

Traveset A, Nogales M, Alcover JA, Delgado JD, Lopez-Da-

rias M, Godoy D, Igual JM, Bover P (2009) A review on

the effects of alien rodents in the Balearic (Western

Mediterranean Sea) and Canary Islands (Eastern Atlantic

Ocean). Biol Invas 11:1653–1670

Trevino HS, Skibiel AL, Karels TJ, Dobson FS (2007) Threats

to avifauna on oceanic islands. Conserv Biol 21:125–132

Underwood AJ (1997) Experiments in ecology. Their logical

design and interpretation using analysis of variance.

Cambridge University Press, Cambridge

Walker BH, Stone L, Hendersen L, Verede M (1986) Size

structure analysis of the dominant trees in a South African

savannah. S Afr J Bot 52:397–402

Wiles GJ, Bart J, Beck RE, Aguon CF (2003) Impacts of the

brown tree snake: patterns of decline and species persis-

tence in Guam’s avifauna. Conserv Biol 17:1350–1360

Zar JH (1974) Biostatistical analysis. Prentice-Hall Inc, NJ

Disruption of recruitment in two endemic palms on Lord Howe Island 3361

123

20

Appendix 2 Auld TD, Leishman MR (2015) Ecosystem risk assessment for Gnarled Mossy Cloud Forest, Lord Howe Island, Australia. Austral Ecology 40, 364–372.

Ecosystem risk assessment for Gnarled Mossy CloudForest, Lord Howe Island, Australia

TONY D. AULD1,2* AND MICHELLE R. LEISHMAN3

1Science Division, NSW Office of Environment and Heritage, PO Box 1967, Hurstville, 2220 (Email:tony.auld@environment.nsw.gov.au), 2Centre for Ecosystem Science, University of New SouthWales,Sydney, and 3Department of Biological Sciences, Macquarie University, North Ryde, New SouthWales,Australia

Abstract Gnarled Mossy Cloud Forest is a globally unique ecosystem, combining floristic elements fromAustralia, New Zealand and New Caledonia. It is restricted to a very small area (28 ha) at elevations above 750 mon the summits of two mountains on Lord Howe Island in the Pacific Ocean, approximately 570 km off the eastcoast of Australia. Moisture derived from clouds is a key feature of the ecosystem. We assessed the conservationstatus of this ecosystem using the International Union for the Conservation of Nature Red List criteria forecosystems. There has been no historical clearing of the ecosystem, but declines (with large uncertainty bounds)were estimated for two abiotic variables that are important in maintaining the component species (cloud cover andrainfall). Overall, we found the ecosystem to be Critically Endangered based on a restricted geographic distributioncombined with continuing decline (criterion B1aii, iii, B1b, B1c and B2aii, iii, B2b, B2c). Decline was inferredfrom: a loss of moisture from declining rainfall and cloud cover due to climate change (affecting disturbanceregimes, gap formation and species survival and recruitment); ongoing exotic rat predation on seeds and seedlingsof several sensitive species that are structural components of the ecosystem (affecting survival and recruitment); andthe fact that the ecosystem is considered to exist at only one location. This mirrors similar threats from exoticspecies and climate change to other Pacific island cloud forests. Eradication of rats from Lord Howe Island willreduce the immediate risk to this ecosystem; however, only global mitigation of greenhouse gases could alleviate riskfrom declining cloud cover and moisture availability.

Key words: moisture loss, rat predation, Red List, restricted distribution.

INTRODUCTION

Risk assessment criteria for ecosystems have beendeveloped by the International Union for the Conser-vation of Nature (IUCN) (Keith et al. 2013). Thesecriteria can be used to both identify those ecosystemsmost in danger of being lost and prioritize actions forprotection and conservation management.The criteriahave elements comparable with those used in red listcriteria for species (Keith et al. 2013) and reflectbroadly similar types of criteria used in some nationallisting criteria for ecological communities (e.g. in Aus-tralia, Nicholson et al. 2014).

Ecosystems of oceanic islands comprise manyunique combinations of both flora and fauna(Mueller-Dombois & Fosberg 1998), making thempriorities for conservation management. Isolated cloudforests are scattered across remote islands of thePacific where they form highly diverse plant commu-nities with high levels of endemism (Meyer 2010).While globally, cloud forest is thought to comprise

some 12% of tropical forests (Mulligan & Burke2005), in Australia and on oceanic islands of thePacific, it is very restricted (Mueller-Dombois &Fosberg 1998; Meyer 2010; Scatena et al. 2010). Herewe assess the conservation status of a cloud forestecosystem endemic to Lord Howe Island. Lord HoweIsland is a small (11 km by 0.6–2.8 km) oceanic islandfound approximately 570 km off the east coast of Aus-tralia, 1350 km from New Zealand and 1250 km fromNew Caledonia (Auld & Hutton 2004). The islandvaries in topography because of two steep rising south-ern mountains (up to 895 m a.s.l.) and is consideredto be a remnant of a large shield volcano (McDougallet al. 1981). The climate of Lord Howe Island ishumid-subtropical.

Lord Howe Island has a mix of floristic elements frommainland Australia, New Zealand and New Caledonia(Pickard 1983; Green 1994), and its entire flora andfauna are considered to have reached the island bylong-distance dispersal (Auld & Hutton 2004). Thisdispersal trait, combined with the large topographicrelief, has led to many of the species assemblages on theisland being globally unique. A large proportion of theplant and invertebrate taxa on the island are endemic

*Corresponding author.Accepted for publication September 2014.

Austral Ecology (2015) 40, 364–372

bs_bs_banner

© 2014 The Authors doi:10.1111/aec.12202Austral Ecology © 2014 Ecological Society of Australia

(105 vascular plant species (44% of the total flora ofLord Howe Island), Green 1994; 600 invertebratespecies (approx. 37% of the total invertebrate fauna ofLord Howe Island), Cassis et al. 2003).

There have been a number of studies on the vegeta-tion of Lord Howe Island (Maiden 1898; Oliver 1916;Pickard 1983; Mueller-Dombois & Fosberg 1998).Given that plants provide a significant component ofthe species diversity as well as a habitat resource forother species, we have here interpreted Pickard’s(1983) vegetation communities as ecosystems. The27-ha summit of Mt Gower and the narrow summit ofMt Lidgbird support an endemic cloud forest knownas Gnarled Mossy Cloud Forest (Harris et al. 2005).In this paper, we assess the conservation status of theGnarled Mossy Cloud Forest ecosystem using theIUCN Red List criteria for ecosystems (Keith et al.2013) by addressing reduction in geographic distribu-tion, restricted geographic distribution, environmentaldegradation and disruption of biotic processes. Wedevelop a conceptual model of ecosystem dynamicsthat would lead to ecosystem collapse (sensu Keithet al. 2013) applicable to cloud forests globally butwith special reference to the disruption to biotic orabiotic processes most relevant to Lord Howe Island.We then used this model to quantify measures ofabiotic degradation and disruption of biotic processes.

ECOSYSTEM DESCRIPTION

Classification

Regionally, this ecosystem is classified as a CriticallyEndangered Ecological Community ‘Gnarled MossyCloud Forest on Lord Howe Island’under the New SouthWales Threatened Species Conservation Act 1995. Theecosystem is currently recognized following the vegetationclassification on Lord Howe Island by Pickard (1983) whodescribed it as Gnarled Mossy Forest. Other studiesdescribe the same ecosystem as Moss Forest (Oliver1916), Cloud Forest (Mueller-Dombois & Fosberg 1998)and Mossy Cloud Forest (Harris et al. 2005).Recent work(Harris et al. 2005) has detailed the plant species compo-sition and internal variation within the ecosystem, alongwith its conservation significance.Under the IUCN Habi-tats Classification Scheme (Version 3.1), the ecosystem isclassified as ‘1. Forest/1.9 Subtropical/Tropical MoistMontane Forest’.

Characteristic native biota

Gnarled Mossy Cloud Forest on Lord Howe Island isa closed-canopy forest of 2–8 m height (Figs 1, 2),with its composition depending on aspect and whether

it occurs on ridges or in drainage lines (Harriset al. 2005). The ecosystem occurs on the summitsof Mt Gower and Mt Lidgbird (Fig. 3). On thesummit of Mt Gower, the dominant plant species areZygogynum howeanum and Dracophyllum fitzgeraldii(Pickard 1983; Harris et al. 2005). Associated treesinclude Cryptocarya gregsonii, Elaeocarpus costatus,Leptospermum polygalifolium ssp. howense, Negriarhabdothamnoides, Pittosporum erioloma, Symplocus can-delabrum, and the palms Hedyscepe canterburyanaand Lepidorrhachis mooreana. Tree ferns (Cyatheaspp.), large tussock sedges (Machaerina insularis andGahnia howeana), ferns Blechnum fullagarii, Blechnumcontiguum, Blechnum howeanum, Grammitis wattsiand other ferns, mosses (e.g. Echinodium hispidum,Spiridens vieillardii; Brotherus & Watts 1915) andlichens (e.g. Strigula spp., Byssoloma subdiscorans,Echinoplaca spp., Gyalectidium filicinum, Porina rufula

Fig. 1. Gnarled Mossy Cloud Forest ecosystem (photo T.Auld).

Fig. 2. Summit plateau of Mt Gower with Gnarled MossyCloud Forest. The very narrow ridgetop of Mt Lidgbird canbe seen at the back, left-hand side (photo T. Auld).

CLOUD FOREST LHI 365

© 2014 The Authors doi:10.1111/aec.12202Austral Ecology © 2014 Ecological Society of Australia

and Trichothelium javanicum; McCarthy et al. 1996)are abundant. Gnarled Mossy Cloud Forest on thesummit ridgetop of Mt Lidgbird above 750 m eleva-tion is much more exposed and restricted in area com-pared with the ecosystem on Mt Gower (Pickard1983). A vegetation plot (10 m by 40 m) on thesummit of Mt Lidgbird in Gnarled Mossy CloudForest had a dominant canopy of H. canterburyana,C. gregsonii, Dysoxylum pachyphyllum, Negria rabdo-thamnoides, P. erioloma and Cyathea macarthurii, alongwith Grammitis diminuta, Carex sp., Olearia mooneyi,Rapanea myrtillina, Z. howeanum, Lordhowea insularis,Gahnia howeana, Coprosma lanceolaris, Dendrobiummoorei, Coprosma putida, Macropiper hooglandii,Microsorium scandens, Asplenium milnei, Aspleniumsurragatum, Elatostema grande, Hymenophyllum sp.(Hutton & Auld, unpubl. data, 2007).

Gnarled Mossy Cloud Forest on Lord Howe Island onlyoccurs on the summit plateau and ridgetops of Mt Gowerand Mt Lidgbird, and is readily distinguished from adja-cent ecosystems that lack the extensive development ofnon-vascular epiphytes and several endemic mountain

vascular plant species.Overall, some 86% of vascular plantspecies in theecosystemareendemic toLordHoweIsland,and some 17% are endemic to the mountain summits onthe island (Harris et al. 2005). Dracophyllum-Metrosiderosscrub (of Pickard 1983 and Mueller-Dombois & Fosberg1998) occurs on benches below the summits. Pickard(1983) notes that Gnarled Mossy Cloud Forest on LordHowe Island is floristically distinct but close toDracophyllum-Metrosideros scrub and H. canterburyanaclosed forest, neither of which are considered to be part ofthe Gnarled Mossy Cloud Forest ecosystem.

Gnarled Mossy Cloud Forest on Lord Howe Island isa key component contributing to the southern moun-tains biodiversity hotspot on Lord Howe Island(Department of Environment and Climate Change2007), particularly for plants and invertebrates. Cassiset al. (2003) found that the assemblage of terrestrialinvertebrates in the Gnarled Mossy Cloud Forestexhibits high species richness, high levels of endemismto Lord Howe Island and many species that arerestricted to the Gnarled Mossy Cloud Forest ecosys-tem. It also forms important habitat for several verte-brates (DECC 2007) including the main breedinghabitat for providence petrels (Pterodroma solandri) andrefugial breeding habitat for the Lord Howe Islandwoodhen (Tricholimnas sylverstis).

Abiotic environment

Lord Howe Island as a remnant volcanic shield isgeologically and topographically diverse (McDougallet al. 1981). The climate is temperate, and sea levelparts of the island have a mean annual temperatureof 19.2°C, ranging from 17°C–25°C in summer to14°C–18°C in winter (Mueller-Dombois & Fosberg1998). At sea level, average annual rainfall is1717 mm, with a maximum of 2886 mm and aminimum of 998 mm (Mueller-Dombois & Fosberg1998). Temperature decreases with altitude in thesouthern mountains (0.9°C for every 100 m rise inaltitude; Simmons et al. 2012). Cloud forests onPacific islands typically occur between 800 and 900 ma.s.l (Meyer 2010), and on Lord Howe Island, theGnarled Mossy Cloud Forest ecosystem occurs from750 to 875 m a.s.l. The annual rainfall in GnarledMossy Cloud Forest is thought to be much higher thanat sea level (although this has not been quantified) andspread throughout the year (DECC 2007). The twosouthern mountains (Mts Gower and Lidgbird) obtainsignificant moisture from both rainfall and directcanopy interception of cloud water (horizontal precipi-tation or cloud stripping), and their peaks are oftenshrouded in cloud (Auld & Hutton 2004). Cloudforests are characterized by increased rainfall andcooler temperatures than forest with no cloud (Jarvis &Mulligan 2011), and this is thought to also apply to theGnarled Mossy Cloud Forest ecosystem.

0 2 4 Kilometres

N

Fig. 3. Map of the two occurrences (shown as the shadedareas) of Gnarled Mossy Cloud Forest on Lord Howe Island.As existing threats are known to occur across both occur-rences, it is considered that these occurrences constitute onelocation under the definitions of the International Union forthe Conservation of Nature (IUCN) guidelines (Keith et al.2013). Source of map data: Pickard (1983).

366 T. D. AULD AND M. R. LEISHMAN

© 2014 The Authorsdoi:10.1111/aec.12202Austral Ecology © 2014 Ecological Society of Australia

Distribution

Gnarled Mossy Cloud Forest is confined to LordHowe Island in New South Wales, Australia. On theisland, it is restricted to the summit plateau of MtGower (some 27 ha) and in a greatly reduced formand extent on the narrow summit ridge of Mt Lidgbird(estimated to be less than 1 ha) (Pickard 1983)(Figs 2, 3). Gnarled Mossy Cloud Forest occurs above750 m in elevation (Pickard 1983).

Key processes, interactions and threats

The ecosystem is sustained by a combination of cloudcover allowing moisture stripping by component species,relatively cool temperatures at >750 m elevation and anocean-moderated climate. This cloud layer provides asource of precipitation (occult precipitation) and main-tains the humidity required by about 86% of the island’sendemic plant species including the profusion of epiphytesin the Gnarled Mossy Cloud Forest (Australian NationalUniversity 2009). Ecosystem dynamics that could lead tocollapse of the Gnarled Mossy Cloud Forest depend ontwo key ecological processes.These are likely increases infrequencyandextentofdisturbance regimesunderclimatechange (e.g. through increased storm damage) and areduction of survival and recruitment of componentspecies in the ecosystem (e.g. through warming, desicca-tion and exotic species) (Fig. 4). Both factors will operateglobally for cloud forest ecosystems (along with the majorissue of loss of habitat), while local factors may also play arole (Fig. 4). As an example, increased desiccation isexpected to reduce species survival and limit opportunities

for recruitment of new individuals.This may result fromcombinations of reduced cloud cover or precipitation,increased warming or increased extreme storm activitythat opens up the existing canopy. In particular, epiphytesmay be very sensitive to changes in microclimate resultingfrom predicted global climate changes (Benzing 1998;Nadkarni 2010), yet these plants play a key role in light,hydrological and nutrient regimes in the forests in whichthey occur (Foster 2001).

Survival and recruitment of species will also beimpacted by local factors, including for Lord HoweIsland, the impact of exotic species such as rats, patho-gens or weeds (Auld & Hutton 2004). Impacts of exoticspecies will also be a global concern for cloud forestsand have been shown to be a major threat to other cloudforests of the Pacific (Meyer 2010). Exotics can selec-tively alter species composition, modify habitats and, incombination with changing moisture availability andwarming temperatures,modify both the structure of theecosystem and the species that can occur in the area.Increased storm damage, declining cloud cover,warming temperatures and impacts of exotic specieswill reduce survival and recruitment in componentspecies in the ecosystem.This is likely to result in botha changed ecosystem composition and ultimately eco-system collapse (Fig. 4).

Globally cloud forest ecosystems are threatened bychanges to land use through clearing, alien species andclimate change (Scatena et al. 2010). On Lord HoweIsland, there are a number of threats to the GnarledMossy Cloud Forest ecosystem (NSW ScientificCommittee 2011) (Fig. 4). Climate change is consid-ered to be a major threat to the ecosystem. As theGnarled Mossy Cloud Forest is confined to mountainsummits, there is no bioclimatic zone for componentspecies to move into with warming temperatures(Australian National University 2009; NSW ScientificCommittee 2011). Decreases in cloud formation(timing, duration and frequency) on the mountains willimpact on the moisture regime and species survival inthe cloud forest (Auld & Hutton 2004) and may favourspecies from lower down the mountains that can bettertolerate desiccation. The ‘lift-cloud-base hypothesis’states that the climate of tropical mountains will gradu-ally change because of an elevation in sea surface tem-peratures (Pounds et al. 1997, 1999; Still et al. 1999).Still et al. (1999) suggest that under a changing climatethere are likely to be upward altitudinal shifts in theareas that experience cloud formations. Pounds et al.(1999) suggest that such a change may have contrib-uted to frog and toad decline in a Costa Rican CloudForest. If this prediction holds for Lord Howe Island,then there may be reduced formation of clouds on thesouthern mountains and reduced moisture availability.The impacts of a rise of the cloud layer, caused by risingsea surface temperatures, constitute a major climate-related threat to the island’s ecosystems. Simmons et al.

Fig. 4. Conceptual model of ecosystem dynamics leadingto collapse for Gnarled Mossy Cloud Forest. Grey boxesrepresent global threats and outcomes for cloud forests,excluding loss of habitat. Unfilled boxes are local threats forLord Howe Island (but may also be applicable to other cloudforests). Arrows indicate direction of transition betweenstates.

CLOUD FOREST LHI 367

© 2014 The Authors doi:10.1111/aec.12202Austral Ecology © 2014 Ecological Society of Australia

(2012) estimated that sea level temperatures aroundLord Howe Island have risen by some 0.6°C since1940. Predicted changes in global mean air tempera-tures from global circulation models forecast a changeof 1.5–6°C by 2100 (Intergovernmental Panel onClimate Change 2007). Australian National University(2009) suggests that average annual air temperature onLord Howe Island is expected to rise (compared with1990 levels) by 1.3 ± 0.6°C by 2030, although there ismuch uncertainty around such estimates and attempt-ing to apply global projections to Lord Howe Island.

The exotic ship rat Rattus rattus has been on LordHowe Island for some 90 years. It is known to havesevere impacts on the two endemic palm genera(Hedyscepe and Lepidorrhachis) that occur in theGnarled Mossy Cloud Forest (Baker & Hutton 2006;Auld et al. 2010), and results in a reduction in groundcover across areas impacted by rats. Rats limit recruit-ment in the palms and consume a large proportion ofseeds and leaves of a number of other taxa (Auld &Hutton 2004; unpubl.). Hence, rats are likely to impacton recruitment in many plant species on Lord HoweIsland. Some 7% of the summit plateau of Mt Gower isbaited to reduce rat impacts, and there is a plan to try toeradicate rats across the whole island (Lord HoweIsland Board 2009). There have been other introduc-tions of exotic animals to Lord Howe Island. Pigs andgoats were naturalized since the mid-1800s. Goats andpossibly pigs had locally significant impacts on GnarledMossy Cloud Forest (Pickard 1983), but the long-termimpacts are unknown. Both have now been eradicatedfrom the island (DECC 2007).

Weeds are only a minor threat on the summit of MtGower, where a small number of weeds have beenrecorded within the Gnarled Mossy Cloud Forest eco-system. Most are facilitated by disturbance and takeadvantage of cliff edges or areas of tree fall, lightningstrike, tree death and other minor disturbances.Theseweeds may compete with native plants for recruitmentin canopy gaps and cliff edges.The species compositionand structure of the ecosystem will be modified if weedsare able to exclude natives.Weeds identified in GnarledMossy Cloud Forest on Mt Gower include: Agrostisavenaceus, Arenaria serpyllifolia, Avena barbata, Bromuscarthartius, Ehrharta erecta, Lolium perenne, Luzulalongiflora, Paspalum sp., Poa annua, Polycarpontetraphyllum, Potentilla indica, Pseudognaphalium luteo-album, Rumex sp., Solanum nigrum, Sonchus oleraceus,Sporobolus africanus, Stellaria media and Vulpiabromoides. On Mt Lidgbird, the more limited extent ofGnarled Mossy Cloud Forest along a narrow ridge(Figs. 2, 3) has allowed more aggressive weedsAgeratina adenophora, Lilium formosanum and S. nigrumto invade (NSW Scientific Committee 2011).

Several other potential threats could impact on speciescomposition and structure in the Gnarled Mossy CloudForest. The exotic pathogen Phytophthora cinnamomi

occurs on Lord Howe Island and could spread to theGnarled Mossy Cloud Forest via walkers or birds,although education measures have been undertaken tominimize this risk. Currently, there are few data on thesusceptibility of key Lord Howe Island endemic plants tothis pathogen (see Auld & Hutton 2004). A number ofpotentially susceptible species are major structural domi-nants in the Gnarled Mossy Cloud Forest. Species occur-ring as part of the Gnarled Mossy Cloud Forest on LordHowe Island and belonging to the Family Myrtaceae(Metrosideros nervulosa and Leptospermum polygalifoliumssp. howense) may be susceptible to exotic rust pathogensincluding myrtle rust, which is present on mainland NSW(NSW Scientific Committee 2011).

METHODS

To assess the Gnarled Mossy Cloud Forest Ecosystem againstthe IUCN Red List for ecosystems criteria (Keith et al. 2013),we searched the available literature for evidence of any threatsto the ecosystem that may result in ecosystem decline.We alsoexamined geographic distribution and both biotic and abioticfactors associated with the ecosystem in the light of theconceptual model of ecosystem dynamics that could lead tocollapse for the Gnarled Mossy Cloud Forest (Fig. 4).

For decline, we examined historical vegetation reports(Maiden 1898, 1914; Oliver 1916; Pickard 1983; Green1994; Mueller-Dombois & Fosberg 1998; Harris et al. 2005).For more recent changes, we inspected satellite imagery(Google Earth and 2011 air photo imagery). We used avail-able Geographical Information System data layers for thedistribution of Gnarled Mossy Cloud Forest (mapped fromPickard 1983) to estimate both the extent of occurrence andarea of occupancy of the ecosystem. Extent of occurrencewas estimated using a minimum convex polygon enclosingthis distributional layer, while area of occupancy was esti-mated by determining the number of 10 × 10 km grids thatthe ecosystem covers.

To assess changes in abiotic features that directly impacton Gnarled Mossy Cloud Forest, we used two metrics basedon collated data from the Australian Bureau of Meteorologyfrom mid 1940s onwards. These metrics relate to how cli-matic moisture conditions may have changed (and the poten-tial for future change). The two metrics were:1. Frequency of cloud cover on the summits of the south-

ern mountains where the Gnarled Mossy Cloud Forestecosystem occurs. We estimated the number of days perannum where there was cloud cover on the summit ofthe two southern mountains of Lord Howe Island usingdata from 1946 to 2013. These peaks are observablefrom the Lord Howe Island meteorological station andcloud cover is recorded several times per day. As theGnarled Mossy Cloud Forest ecosystem is restricted toelevations above 750 m, this metric allowed us to esti-mate the long-term changes in cloud cover directly rel-evant to this ecosystem. These data were used tocalculate changes in the number of days of cloud cover(with plausible bounds) over the last 50 years and for thenext 50 years. In cloud forests, the degree of cloud covervaries with elevation and distance from the coast (Jarvis

368 T. D. AULD AND M. R. LEISHMAN

© 2014 The Authorsdoi:10.1111/aec.12202Austral Ecology © 2014 Ecological Society of Australia

& Mulligan 2011). As cloud forests are confined to habi-tats with generally >60% cloudy days per annum(Mulligan & Burke 2005) and data from Lord HoweIsland suggested cloud cover was historically in thisrange (average over 78 years of 69%, annual range of55–83%), we chose a threshold of collapse of 50%cloudy days per year for this index.

2. Changes to rainfall. The meteorological station, locatedjust above sea level, can be used to estimate if there hasbeen any decline in rainfall. We used data on annualrainfall for Lord Howe Island that had been analysed bySimmons et al. (2012) to make inferences about chang-ing rainfall on the island that would also reflect changesto moisture availability in the Gnarled Mossy CloudForest ecosystem. Use of this index assumes thatchanges in rainfall at sea level would indicate a compara-tive level of change at the mountain top elevations ofGnarled Mossy Cloud Forest and that any decreases willreduce the persistence of species within the ecosystem.Again, comparisons were made between annual rainfallmeasures in the last 50 years and for the next 50 years.For rainfall, Pickard (1983) reported dying plant frondsand leaf fall on lowland trees and desiccation of herbs onLord Howe Island during a dry period in 1977. Annualrainfall on Lord Howe Island in 1977 was 1071 mm,compared with the average of 1526 mm. 1977 was thesecond lowest annual rainfall recorded in the last 73years (the lowest being in 1997 at 990 mm). Hence, weused a collapse threshold of 900 mm annual rainfall(below the lowest ever recorded).

Temperature increases have been identified for near sea levelon Lord Howe Island (Simmons et al. 2012),and any increases intemperatures observed at sea level are likely to be reflected intemperatures on the southern mountains as Simmons et al.(2012) found an adiabatic lapse rate of some 0.9°C per 100 mincrease in altitude on Mt Gower on Lord Howe Island.We didnot use temperature as an abiotic variable as we had no data toinform the threshold of collapse of Gnarled Mossy Cloud Forestin relation to temperature. Instead, we examined the strength ofthe correlations between annual average minimum andmaximum temperatures with both cloud cover frequency andannual rainfall to determine if temperature might be a usefulpredictor of these abiotic variables.

There is further potential abiotic degradation from anyincreases in storm damage from extreme events underclimate change. Remote sensing data could also be used toestimate changes in storm frequency; however, no index ofthis is currently available.

For biotic degradation, we reviewed the impact of exotic ratson limiting recruitment of key species in the ecosystem and inchanging the structure and composition of the ecosystem.

For assessment of criteria A and B, collapse was assumedto occur when the mapped distribution of the ecosystemdeclines to zero. For criterion C, ecosystem collapse wasdefined in relation to changes in cloud days (collapse at 50%)or precipitation (collapse at annual rainfall of 900 mm).Decreases in cloud days or precipitation lead to a reductionin the persistence of species within the ecosystem until thethresholds of collapse are reached. Under criterion D, eco-system collapse was assessed based on changes in communitycomposition, particularly loss of canopy dominants and char-acteristic epiphytes.We could not undertake an estimation of

the probability of ecosystem collapse (criterion E), and thiscriterion was considered to be Data Deficient.

RESULTS

Criterion A – decline in distribution

There have been no observed changes in the extent of thisecosystem in the last 50 years (Pickard 1983).The remotelocation, steep terrain and conservation value of thesouthern mountains has ensured no clearing of theecosystem. No changes are indicated since Pickard’s 1983vegetation classification based on inspection of recent sat-ellite imagery (Google Earth and 2011 air photo imagery)and field traverses.The status of the ecosystem is thereforeLeast Concern under criterion A1.

No projections are available for the future distribu-tion, although it is possible the current bioclimaticenvelope for the ecosystem may shift sufficiently tocause ecosystem collapse. Further data on climaticconditions within the ecosystem are needed toexamine this possibility.The status of the ecosystem istherefore Data Deficient under criterion A2.

There has been no observed change in the extent ofthis ecosystem since 1750 (Pickard 1983). The statusof the ecosystem is therefore Least Concern undercriterion A3.

Criterion B – distribution size

The extent of occurrence was estimated to be 20 km2

based on a minimum convex polygon enclosing alloccurrences mapped by Pickard (1983).

The area of occupancy of Gnarled Mossy CloudForest is estimated to be one 10 × 10 km grid cell.

Long-term monitoring data on both rainfall andcloud cover provide evidence for a continuing declinein abiotic components of the ecosystem (AustralianNational University 2009, Simmons et al. 2012, seebelow and Fig. 5). Additionally, rat predation on seedsand seedlings of several sensitive species (e.g. moun-tain palms) that are structural components of the eco-system is inferred to result in continuing decline of thestanding vegetation and seed banks (Auld & Hutton2004; Auld et al. 2010). This rat predation on seedsand seedlings is likely to cause continuing reduction/elimination of recruitment of several key canopyspecies in the next 20 years. The distribution of thisecosystem (Mt Gower and Mt Lidgbird) is consideredto be a single location, as the existing known threatsoccur across both remaining patches.This renders theecosystem susceptible to stochastic processes such astropical cyclones and wildfire within a short timeperiod in an uncertain future.

CLOUD FOREST LHI 369

© 2014 The Authors doi:10.1111/aec.12202Austral Ecology © 2014 Ecological Society of Australia

The status of the ecosystem is therefore CriticallyEndangered under criteria B1aii, iii, B1b, B1c andB2aii, iii, B2b, B2c, and Vulnerable under B3 (seeTable 1).

Criterion C – decline in abiotic processesand features

There was a trend for decline in the two moisture indicesused. The extent of impact extends across 100% of theecosystem for both indices. For cloud cover, ecosystemcollapse is suggested to occur when there is cloud coverfor less than 50% of the year (Mulligan & Burke 2005). InCosta Rica, Pounds et al. (1999) showed that increases indry days (and hence a decline in mist frequency) wasassociated with increases in bird species from habitatsbelow cloud forests, along with declines in populations of

cloud forest anoline lizards and anurans. Cloud cover overthe last 50 years declined an estimated 22% (95% Con-fidence Limit 4–38%) towards the threshold of collapseset at 50% cloudy days per year (R2 = 8.1%, Fig. 5).Thiswould imply a best estimate of least concern, but the 95%confidence limits of the estimate would include LeastConcern to Vulnerable.

For annual rainfall, we found over the last 50 yearsthere had been a decline of 31% (95% CL 4–79%)towards the collapse threshold. This would imply abest estimate of Vulnerable, but the 95% confidencelimits of the estimate would include Least Concern toEndangered. Hence, the status of the ecosystem undercriterion C1 is Least Concern to Vulnerable.

Both minimum and maximum temperatures at sea levelincreased in the last 50 years (Simmons et al. 2012).Thelong-term average maximum and minimum tempera-tures are 21.87°C and 16.93°C, respectively.The highestaverage yearly maximum temperature (22.6°C) was in1973, and 4 of the warmest 5 years have been since 1998.We found poor correlations between annual rainfalland the proportion of cloudy days or temperature. Thehighest correlations were between cloud cover and averageannual minimum temperature (–0.4) and average annualmaximum temperature (–0.31), suggesting that increasesin these temperature variables could be a factor in influ-encing cloud cover changes.

The status of the ecosystem under criterion C2 canbe assessed by extrapolating the rate of change incloud cover or rainfall over the next 50 years againstthe thresholds for collapse. This suggested that cloudcover would be reduced by 28% (95% CL range4–60%) and annual rainfall by 44% (95% CL range4–100%) relative to collapse thresholds over the next50 years. Consequently, the status of the ecosystemunder C2 would be predicted to be Vulnerable toEndangered (with a plausible range of Least Concernto Collapsed). The status of the ecosystem under cri-terion C3 (Historic decline) is Data Deficient.

Criterion D – decline in biotic processesand interactions

The ecosystem is subject to the impact of exotic rats.Rats damage key functional plant species in the eco-system via predation on seeds and seedlings (Auld &Hutton 2004; Auld et al. 2010). Most of the area(93%) of Gnarled Mossy Cloud Forest is subject toimpact by rats (7% is baited to reduce impact). Theseverity of impacts is not known for all speciesimpacted, but for two palm species (Auld et al. 2010)it varies from 100% (ongoing recruitment failure) to>75% (recruitment failure at 75% of surveyed sitesand significantly reduced at 25% of sites). Thus, theextent and severity of seed predation by rats is high fortwo key species, but the severity of seed predation byrats on other taxa in the ecosystem is uncertain.

19401950

19601970

19801990

20002010

2020

Year

0.5

0.6

0.7

0.8

0.9P

ropo

rtio

n of

clo

ud c

over

day

s pe

r ye

ar

Fig. 5. Estimated change in the proportion of days per yearwith cloud cover in Gnarled Mossy Cloud Forest over 1945to 2013. Regression is proportion cloudy days = 2.50 –0.00091 × year. The standard error of the slope is 0.00038.

Table 1. Summary of the conservation assessment forGnarled Mossy Cloud Forest against the IUCN criteria forecosystems

Criterion A B C D E Overall

Subcriterion 1 LC CR LC-VU DD DD CRSubcriterion 2 DD CR VU-EN DDSubcriterion 3 LC VU DD DD

CR, critical endangered; DD, data deficient; EN, endan-gered; IUCN, International Union for the Conservation ofNature; LC, least concern; VU, vulnerable.

370 T. D. AULD AND M. R. LEISHMAN

© 2014 The Authorsdoi:10.1111/aec.12202Austral Ecology © 2014 Ecological Society of Australia

The ecosystem is also susceptible to P. cinnamomi andmyrtle rust as several dominants in the canopy are likely tobe impactedby thesepathogens.Phytophthora cinnamomi ispresent on Lord Howe Island (Auld & Hutton 2004). Noestimate of severity or impact is available.

The status of the ecosystem under criteria D1, D2and D3 are all Data Deficient.

DISCUSSION

Unlike other global cloud forest ecosystems (Scatenaet al. 2010), Gnarled Mossy Cloud Forest has notbeen subject to recent clearing or fragmentation. Wefound Gnarled Mossy Cloud Forest to be CriticallyEndangered based on a restricted geographic distribu-tion combined with continuing decline in abiotic andbiotic processes inferred from: a loss of moisture fromdeclining rainfall and cloud cover due to climatechange; ongoing exotic rat predation on seeds andseedlings of several sensitive species that are structuralcomponents of the ecosystem; and the single locationof the ecosystem. These are likely to lead to reducedsurvival and recruitment in species in the ecosystem(Fig. 4). Hence, this ecosystem is impacted by similarthreats to other Pacific island cloud forests (Meyer2010).

We successfully applied several related indices ofchange in the climate that may affect the GnarledMossy Cloud Forest ecosystem. However, the sensitiv-ity of relationships between the degree of decline ineither cloud cover or rainfall, and species and ecosys-tem decline remains somewhat uncertain. Our esti-mates of thresholds for collapse are based on limitedinformation and the estimates of decline in cloud coverand rainfall had large plausible bounds (e.g. resultingin 95% confidence limit estimates of Least Concern toCollapsed in C2). While cloud cover data have themost direct relevance to species survival in the ecosys-tem, average annual temperature changes were nega-tively correlated with cloud cover suggesting the ‘liftcloud base hypothesis’ (Pounds et al. 1997, 1999; Stillet al. 1999) may be applicable to Lord Howe Island. Itis possible that increases to temperature near sea levelmay be a suitable surrogate for changes in cloud coverin the Gnarled Mossy Cloud Forest ecosystem. Dataon biotic processes and interactions were available foronly two of the canopy palm species in the ecosystem.While we had data on both severity (75–100%) andextent (93%) for these species, uncertainty remainsabout the severity across other key structural taxa inthe ecosystem (just how many species need to beimpacted to drive the ecosystem to decline). As well, asmature plants may be long-lived, recruitment failurewould need to consistently occur across decades (as isthe case in the two palms; Auld et al. 2010). Factorsinfluencing adult tree survival (such as reduced mois-

ture availability, increased extreme weather and light-ning strikes) will also be important estimating theseverity of rat impacts.

Future prognosis for the ecosystem

A rodent eradication programme has been developedfor Lord Howe Island (Lord Howe Island Board 2009;Wilkinson & Priddel 2011) and is planned to occur inthe coming years. If successful, this has the potential toeliminate the major immediate threat to thisecosystem. If unsuccessful, regular rodent baitingacross much of the distribution of the ecosystemwould be needed, although this is not likely to be along-term solution as the rodents gain resistance to thepoisons. A reduction in rat impact is needed to bothallow successful seed production and recruitment ofnew plants. As rats also consume leaves of developingplants, a period of a decade or more of minimal ratimpacts would be needed to promote regeneration.

Phytosanitory guidelines have been developed forLord Howe Island in an effort to reduce any risk ofspread of the existing P. cinnamomi (DECC 2007) andlimit the introduction of novel pathogens such asmyrtle rust. An ongoing programme of early detectionis needed to complement this. Weed impacts are cur-rently minor except for the ecosystem occurrence onMt Ligdbird, where large habitat edges favouringcrofton weed are impacting on the ecosystem. Unfor-tunately, the remote and rugged nature of the terraincurrently prevents any control efforts.

Any prevention of climate warming and reduction incloud-derived precipitation will depend on global effortsto reduce greenhouse gas emissions.The current progno-sis for this is not favourable (Peters et al. 2013).

Collapse of the Gnarled Mossy Cloud Forest Eco-system would have cascade effects for many of thecomponent species and others dependent upon thisecosystem. Visiting the ecosystem is also part of thepopular tourist walk to Mt Gower. Decline and loss ofthis unique ecosystem would likely impact on thepopularity of this walk and the tourism values of theisland.

ACKNOWLEDGEMENT

This work was supported by an Australian ResearchCouncil Linkage Grant (LP LP130100435). JudyScott and Chris Simpson developed the distributionmap of Gnarled Mossy Cloud Forest.We thank DavidKeith for his helpful comments on earlier versionsof this manuscript and all the participants of the Aus-tralian Centre for Ecological Analysis and Synthesisworkshop for their thoughtful contributions to thedevelopment of the risk assessment for this ecosystem.

CLOUD FOREST LHI 371

© 2014 The Authors doi:10.1111/aec.12202Austral Ecology © 2014 Ecological Society of Australia

REFERENCES

Auld T. D. & Hutton I. (2004) Conservation issues for thevascular flora of Lord Howe Island. Cunninghamia 8, 490–500.

Auld T. D., Hutton I., Ooi M. K. J. & Denham A. J. (2010)Invasive species on oceanic islands: disruption of recruit-ment in narrow endemic palms. Biol. Invasions 12, 3351–61.

Australian National University (2009) Implications of climatechange for Australia’s World Heritage properties: a prelimi-nary assessment. A report to the Department of ClimateChange and the Department of the Environment, Water,Heritage and the Arts by the Fenner School of Environmentand Society, the Australian National University.

Baker W. J. & Hutton I. (2006) Lepidorrhachis. Palms 50, 33–8.Benzing D. H. (1998) Vulnerabilities of tropical forests to climate

change: the significance of resident epiphytes. Clim. Change39, 519–40.

Brotherus V. F. & Watts W.W. (1915) The mosses of Lord HoweIsland. Proc. Linn. Soc. N. S.W. 40, 363–85.

Cassis G., Meades L., Harris R. et al. (2003) Lord Howe Islandterrestrial invertebrate biodiversity and conservation.Report to the NSW National Parks and Wildlife Service bythe Australian Museum Centre for Biodiversity and Con-servation Research.

Department of Environment and Climate Change (2007) LordHowe Island biodiversity management plan. Department ofEnvironment and Climate Change (NSW), Sydney.

Foster P. (2001) The potential negative impacts of global climatechange on tropical montane cloud forests. Earth Sci. Rev.55, 73–106.

Green P. S. (1994) Flora of Australia Vol. 49 Oceanic Islands 1.Australian Government Printing Service, Canberra.

Harris R., Cassis G., Auld T. & Hutton I. (2005) Floristics andstructure of the mossy cloud forest of Mt Gower summit,Lord Howe Island. Pac. Conserv. Biol. 11, 246–56.

Intergovernmental Panel on Climate Change (2007) ClimateChange 2007: The Physical Science Basis. Contributions ofWorking Group 1 to the Fourth Assessment Report of the Inter-governmental Panel on Climate Change. Cambridge Univer-sity Press, Cambridge.

Jarvis A. & Mulligan M. (2011) The climate of cloud forests.Hydrol. Process. 25, 327–43.

Keith D. A., Rodríguez J. P., Rodríguez-Clark K. M. et al. (2013)Scientific Foundations for an IUCN Red List of Eco-systems. PLoS ONE 8, e62111.

Lord Howe Island Board (2009) Draft Lord Howe Island rodenteradication plan. Lord Howe Island Board, Lord HoweIsland.

McCarthy P. M., Streimann H. & Elix J. A. (1996) Newfoliicolorus species of Strigula from Lord Howe Island,Australia. Lichenologist 28, 239–44.

McDougall I., Embleton B. J. J. & Stone D. B. (1981) Origin andevolution of Lord Howe Island, southwest Pacific Ocean.J. Geol. Soc. Austr. 28, 155–76.

Maiden J. H. (1898) Observations of the vegetation of LordHowe Island. Proc. Linn. Soc. N. S.W. 23, 112–58.

Maiden J. H. (1914) Further notes on the botany of Lord HoweIsland (fifth paper). Proc. Linn. Soc. N. S.W. 39, 377–84.

Meyer J.-Y. (2010) Montane cloud forests on remote islands ofOceania: the example of French Polynesia (South Pacific

Ocean). In: Tropical Montane Cloud Forests: Science for Con-servation and Management (eds L. A. Bruijnzeel, F. N.Scatena & L. S. Hamilton) pp. 121–9. Cambridge Univer-sity Press, Cambridge.

Mueller-Dombois D. & Fosberg F. R. (1998) Vegetation of theTropical Pacific Islands. Springer-Verlag, New York.

Mulligan M. & Burke S. (2005) Global cloud forests and envi-ronmental change in a hydrological context. DFID FRPProject ZF0216. Final Report. December 2005. [Cited 23September 2014.] Available from URL: http://www.ambiotek.com/cloudforests/cloudforest_finalrep.pdf

Nadkarni N. M. (2010) Potential effects of global climate changeon epiphytes in a tropical montane cloud forest: an experi-mental study from Monteverde, Costa Rica. In: TropicalMontane Cloud Forests: Science for Conservation and Manage-ment (eds L. A. Bruijnzeel, F. N. Scatena & L. S. Hamilton)pp. 557–65. Cambridge University Press, Cambridge.

Nicholson E., Regan T. J., Auld T. D. et al. (2014) Towardsconsistency, rigour and compatibility of risk assessments forecosystems and ecological communities. Austral Ecol. onlineearly.

NSW Scientific Committee (2011) Final determination.Gnarled Mossy Cloud Forest on Lord Howe Island – criti-cally endangered ecological community listing. [Cited 23September 2014.] Available from URL: http://www.environment.nsw.gov.au/determinations/gnarledmossycloudforestlhifd.htm

Oliver W. R. B. (1916) The vegetation and flora of Lord HoweIsland. Trans. Proc. N. Z. Inst. 49, 94–161.

Peters G. P., Andrew R. M., Boden T. et al. (2013) The challengeto keep global warming below 2°C. Nat. Clim. Change 3,4–6.

Pickard J. (1983) Vegetation of Lord Howe Island.Cunninghamia 1, 133–265.

Pounds J. A., Fogden M. P. L. & Campbell J. H. (1999) Biologi-cal response to climate change on a tropical mountain.Nature 398, 691–715.

Pounds J. A., Fogden M. P. L., Savage J. M. & Gorman G. C.(1997) Tests of null models for amphibian declines on atropical mountain. Conserv. Biol. 11, 1307–22.

Scatena F. N., Bruijnzeel L. A., Bubb P. & Das S. (2010) Settingthe stage. In: Tropical Montane Cloud Forests: Science for Con-servation and Management (eds L. A. Bruijnzeel, F. N.Scatena & L. S. Hamilton) pp. 3–13. Cambridge UniversityPress, Cambridge.

Simmons L., Auld T. D., Hutton I., Baker W. J. & Shapcott A.(2012) Will climate change, genetic and demographic vari-ation or rat predation pose the greatest risk for persistence ofan altitudinally distributed island endemic? Biology 1, 736–65.

Still C. J., Foster P. N. & Schneider S. H. (1999) Simulating theeffects of climate change on tropical montane cloud forests.Nature 398, 608–10.

Wilkinson I. S. & Priddel D. (2011) Rodent eradication on LordHowe Island: challenges posed by people, livestock, andthreatened endemics. In: Island Invasives: Eradication andManagement (eds C. R. Veitch, M. N. Clout & D. R. Towns)pp. 508–14. IUCN, Gland.

372 T. D. AULD AND M. R. LEISHMAN

© 2014 The Authorsdoi:10.1111/aec.12202Austral Ecology © 2014 Ecological Society of Australia