Hilton 2006
New Zealand Geographer (2006) 62, 105–120 10.1111/j.1745-7939.2006.00054.x
Research Article
Original Article
The spread of marram grass
Blackwell Publishing Asia in NZ
The loss of New Zealand’s active dunes and the
spread of marram grass (Ammophila arenaria)
Michael J. Hilton
Department of Geography, University of Otago, PO Box 56, Dunedin, New
Zealand
Abstract: This article examines the decline of New Zealand’s active dunes in relation to
the introduction of marram grass (Ammophila arenaria). The area of active dunes in New
Zealand declined from 129 000 ha in the early 1900s to about 39 000 ha in 2000; a reduction
of 70%. The extent of active dunes has declined since the 1950s in all regions, particularly
in Northland, Auckland and the Manawatu. The loss of active dunes on the west coast of
the North Island resulted primarily from the introduction of marram grass, followed by the
establishment of Pinus radiata plantations and extensive pastoral farming. Between 1985
and 2005 marram grass extended its range to the detriment of the indigenous foredune
flora. Conservation and resource management agencies should urgently identify dune systems
for conservation management and marram grass eradication.
Key words: afforestation, dune conservation, dune inventory, marram grass.
Coastal dunes of Late-Holocene age are a one widespread strandline species (the recently
distinctive element of the New Zealand coast. introduced Cakile edentula) and only three
Active dunes, the focus of the present paper, indigenous, primary sand-colonizing, foredune
are characterized by ongoing eolian sedimenta- species – Desmoschoenus spiralis (a sedge),
tion and a sparse or patchy vegetation cover. Spinifex sericeus (a grass) and Austrofestuca
They are readily distinguished from older and littoralis (a grass). Compared with European
stable dunes, associated with earlier phases of dunes, the active dunes of New Zealand, in-
dune development (Shepherd 1987; Hesp 2001). cluding the contemporary foredunes, are sparsely
These older dunes were, in the main, forested vegetated. In particular, New Zealand dunes
before European settlement and show advanced lack the diversity of grass species of the
podzol development (Cowie 1963). The contem- genera Ammophila, Leymus, and Elymus, which
porary active dunes of New Zealand contain are widespread on the foredunes of Europe
a diversity of dune environments and plant (Doody 1991, 2001).
communities (Johnson 1992), many of which The active dunes of New Zealand represent
are local equivalents of the European dune the most recent phase of dune mobility and
ecosystem complexes identified by Olson and dunefield formation. The genesis of these
van der Maarel (1989). In other respects, the dunefields has been examined, but not resolved.
active dunes of New Zealand are very different Debate has centred on the relative importance
from European dunes. New Zealand has only of human versus climatic disturbance (McGlone
Note about the author: Mike Hilton is a senior lecturer in the Department of Geography, University of Otago. His current
research is concerned with the interpretation, management and restoration of coastal sand systems (particular
dunes); the dispersal and invasion ecology of exotic dune plants; and methods of coastal management in New
Zealand and Southeast Asia. He is an advocate for the conservation and sustainable management of coastal
resources in relation to a range of coastal resources issues, particularly the mining of coastal sand systems and the
use and restoration of active coastal dunelands.
E-mail: mjh@geography.otago.ac.nz
© 2006 The Author
Journal compilation © 2006 The New Zealand Geographical Society Inc. Published by Blackwell Publishers Ltd.
106 M. J. Hilton
1983; McFadgen 1989). Human use of New Previous inventories
Zealand dunes is very recent compared with
the long history of occupation in Europe. The Several government agencies have prepared
Polynesian settlers of New Zealand may well inventories of ‘coastal dunes’ (Table 1). The
have disturbed, de-vegetated and destabilized purpose and methodology of these surveys
particular dune systems. It is clear, however, has not been consistent. The first survey was
that active dunes have been heavily modified undertaken by Leonard Cockayne in the early
since the arrival of Europeans in New Zealand, 1900s. Cockayne (1911) estimated the total area
primarily as a result of the recent, widespread of ‘dune’ in 1911 to be 128 740 ha, of which
and very rapid stabilization of active dune systems 118 900 ha were located in the North Island
with marram grass (Ammophila arenaria). and 9 840 ha in the South Island and Stewart
The majority of active dunes in New Zealand Island. These estimates excluded established
now bear little resemblance to the accounts shrub and forest on stabilized dunes, but
of Leonard Cockayne, one of New Zealand’s included wetlands within active dune systems.
founding botanists. Since Cockayne’s early The same approach has been adopted in the
descriptions of the flora and vegetation com- present study.
munities of dunes (Cockayne 1909, 1911), The New Zealand Land Resource Inventory
marram grass has been planted throughout (NZLRI) mapped the area of active dunes
New Zealand to stabilize active dunes, establish at a scale of 1 : 1 000 000 (Newsome 1987).
foredunes for property protection and prepare For that purpose, sand dune communities were
dunes for afforestation with North American defined as ‘communities dominated by herba-
conifers, particularly Pinus radiata . These ceous plants and low shrubs occurring on recent,
activities, often in conjunction with sand mining, unstable dune sands’ (Newsome 1987, p. 130).
stock grazing, infrastructure development, Stable backdune habitat, typically dominated
urbanization and the introduction of a wide by woody shrub and tree species, was excluded.
range of exotic plant, shrub and tree species Because of the map scale employed, many
(e.g. Acacia sophorae and Ehrharta villosa smaller active dune systems, particularly along
var. maxima), have contributed to the loss of the east coast of New Zealand, were omitted.
the extent and natural character of active dune The maps of active dune systems were derived
systems. Coastal sand dunes may be New from data collected during the 1960s and
Zealand’s most threatened natural habitat, 1970s for the NZLRI (NWASCO 1975–1979).
measured both in terms of the diminished The total area of active dune corresponding to
extent and modification of vegetation of Newsome’s definition was estimated at 52 000 ha.
remaining areas. A further 40 000 ha of backdune, dominated
New Zealand is a signatory to Agenda 21 by exotic lupins (Lupinus arboreus) and shrubs,
and the Biodiversity Convention. Signatories were identified. In addition, approximately
commit to preparing inventories of critical 200 000 ha of ‘former dunes’ were described as
coastal habitats, conserving and restoring being covered in pasture grasses, pine trees,
critical habitats and identifying critical areas. gorse (Ulex europeaus) and other exotic species
In relation to coastal dunes, progress has (Newsome 1987). Of the 305 000 ha of coastal
been mixed and information on the conserva- sand dunes identified in the NZLRI (NWASCO
tion status of active dunes is incomplete. The 1975–1979), less than 10% were considered
present study aims to (1) identify and map close to their ‘original condition’ (Hunter &
active dunes with the intention of producing a Blaschke 1986). The phrase ‘original condition’
comprehensive national inventory; (2) reveal is presumed here to mean their condition
trends in the reduction of active dune systems around 1900, prior to the widespread stabiliza-
in each region; and (3) document the recent tion of active dunes with marram grass.
spread of marram grass since the first national A national programme to survey the diversity
census of dune flora in the early to mid-1980s. of New Zealand’s indigenous flora, landforms
The strengths and weaknesses of existing dune and landscapes was initiated in 1983. The
inventories are discussed and further know- Protected Natural Areas Programme (PNAP)
ledge requirements identified. aimed to record the location and characteristics
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
Table 1 Comparison of past dune inventories with the present study (LRI – Land Resource Inventory; PNAP – Protected Natural Areas Programme;
LENZ – Land Environments of New Zealand)
Area (ha)
Mapping South Island & National
Authority/Study Source (date) Scale North Island Stewart Island total Comments:
Cockayne (1911) C/F unknown 118 900 9840 128 740 Cockayne included active dunes, but excluded shrubland and forest on
stabilized dunes
Land Resource V (1960/70s) 1 : 1 000 000 52 000 Excludes 40 000 ha of ‘backdune’ dominated by exotic shrubs and 200 000 ha
Inventory of former dunes under pasture and plantation forest (Newsome 1987).
The spread of marram grass in NZ
Protected Natural 1983– 1 : 50 000 Involves sub-regional surveys of ‘ecological districts’ with the objective of
Areas Programme identifying RAPs (recommended areas for protection). Incomplete national
cover; most districts containing dunes have not yet been surveyed.
Sand Dune & F (1984–1988) no maps Assessment of conservation value of most dune systems in New Zealand;
Beach Vegetation no mapping or area estimates of coastal dunes. This study identified 53
Journal compilation © The New Zealand Geographical Society 2006.
Inventory national priority sites for conservation (Johnson 1992; Partridge 1992).
Land Cover S (1996–1997) 1 : 50 000 30 100 11 900 42 000 Maps active dunes, but excludes older vegetated dunes.
Database
Land Environments V (1998) 1 : 50 000 National classification of ‘environments’ based on variation in climatic,
of New Zealand landform and soil variables.
This study C/A (1950s) 1 : 25 000– 110 884 18 518 129 402 Maps active dunes. Excludes older, forested dunes; also maps active dunes
1 : 63 660 in 1950s, 1960s and 1980s by region.
C/A (1990s) 1 : 50 000 28 763 10 041 38 804
Source of information: V – soil and other environmental data; C – cadastral and land cover maps; F – field survey; A – vertical aerial photography; S – satellite data.
Existing classifications are generally based on physical geography and use manual classification techniques to draw visible ecological boundaries onto maps. Newer classifications
© 2006 The Author
such as LENZ are numerically based using computer programs to sort climate, landform and soils data to group areas containing ecosystems of similar type. Both approaches
provide a wealth of information for conservation and resource management.
The numerically based approach means there are no geographic constraints and the classification can be used at varying scales, i.e. environments (containing similar ecosystems)
can vary markedly in size and can be identified even where they are distributed across the landscape in a number of pieces.
107
108 M. J. Hilton
of indigenous plant communities in the know- or animal modification of the site, and the degree
ledge that many remaining communities of high of invasion by weed species. Some large dune
conservation value were poorly represented systems were not visited (e.g. Kokota Spit in
in the existing system of national parks and Northland) and some dune systems, subsequently
reserves. Recommended Areas for Protection identified as having high conservation values
(RAPs) were assessed and identified on the (e.g. Tokomairiro River in Otago), were over-
basis of representativeness, diversity, special looked (Rapson 1996).
features, naturalness, viability, size and shape, The Inventory identified 23 sites in the North
and buffering from external stresses. The bound- Island, 21 in the South Island, and 9 on Stewart
aries of individual RAPs were delineated on Island as ‘national priority dune systems for
topographic maps at a scale of 1 : 50 000. conservation’. All had scores of 15 or more.
The PNAP has provided valuable information Some regions have no sites of national signi-
on the botanical characteristics and conserva- ficance (e.g. Otago). The majority of recognized
tion status of some dune systems, including sites are in the far north and far south of New
those in the Manawatu region (Ravine 1992). Zealand.
Several RAPs contain active or stabilized dunes. As yet, there is no comprehensive database
Most are fragments of much larger, modified of the extent and conservation values of
dunefields. Relatively few ecological districts New Zealand’s active dune systems. The Sand
have been surveyed since the PNAP began in Dune and Beach Vegetation Inventory, despite
1983. There are 179 ecological districts: 104 its methodological limitations, has provided
in the South Island and offshore islands, and the only consistent, nation-wide, assessment of
75 in the North Island. Of these, 123 contain the conservation status of dune systems. It did
stretches of coast. To date, only 33 PNAP not, however, map the boundaries or extent
reports have been published and only 25 of the of individual dune systems, active or forested.
123 coastal ecological districts have been The PNAP surveys provide more detailed
surveyed (18 in the North Island, 7 in the site information, show dune systems of high
South Island). Ecological districts known to conservation value and recommend sites for
contain dune systems of national significance protection, but to date very few dune systems
have yet to be surveyed. Two clusters of dunes have been surveyed. The current character,
of national significance occur in the far north and change in character, of a handful of dune
of the North Island and far south of the South systems have been described (Walls 1998), but
Island (identified by Johnson 1992 and Partridge trends in the state of New Zealand’s dune
1992). The PNAP has not provided the rapid systems have not been documented. There-
synoptic survey of New Zealand’s vegetation fore, the current study identifies trends in the
cover envisaged in the early 1980s. decline in area of active dunes at regional and
The Sand Dune and Beach Vegetation Inven- national scales.
tory (henceforth ‘the Inventory’) conducted by
the Botany Division of the former Department The nature of ‘active dunes’
of Scientific and Industrial Research (Johnson This study is concerned with ‘active dunes’.
1992; Partridge 1992), provided the first national The contemporary ecology of these dunes is
survey and systematic conservation assessment primarily determined by eolian sedimentation.
of coastal dunes in New Zealand (both active Plant diversity is low and mainly restricted
and forested). The aim of that survey was to to species adapted to, or tolerant of, erosion,
identify dune systems with high natural values burial, extreme solar radiation, desiccation and
for conservation purposes (with an emphasis salt spray. Desmoschoenus spiralis (pingao),
on botanical values). Fieldwork was carried out Austrofestuca littoralis (sand tussock), Spinifex
between 1984 and 1988 by regional botanists sericeus, Ammophila arenaria (marram grass)
of the Department of Scientific and Industrial and Euphorbia glauca (shore spurge) are
Research. Individual dune systems were scored commonly associated with active dunes. They
0 to 20, based on the diversity of plant communities tend to form relatively dense stands across
and landforms, the number or proportion of foredunes, but cover may be sparse or patchy
native sand dune species, the degree of human within larger dune systems associated with
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 109
transgressive dunes. A collection of other proximity to the foreshore; and (4) morpho-
species, including the sand daphne (Pimelea logical evidence of transgressive dune forma-
spp.), sand coprosma (Coprosma acerosa), and tion and migration. Two cases, Karikari Beach
Isolepis nodosus, are commonly associated with in the Northland Region, and Mason Bay
active dunes. on Stewart Island (Southland), illustrate how
Active dunes are usually associated with the these criteria were applied. The active dunefield
most recent phase of dunefield development east of Puheke, a conspicuous volcanic cone,
around the New Zealand coast – equivalent to comprises a foredune (dominated by Desmo-
the Hoatan (McFadgen 1989) or the Waitarere schoenus), backed by a sparsely vegetated
phase (Cowie 1963; Muckersie & Shepherd 1995). dunefield that contains multiple transverse
They are readily distinguished from earlier dunes (Fig. 1). The seaward edge of the dune
phases of dune development through soil system is defined by the toe of the vegetated
development, in that they show virtually no foredune. The landward boundary is defined
modification of the grey, quartzo-feldspathic by a wetland. Older dunes, probably of last
sand parent material. This most recent phase interglacial age, form the hinterland. The second
is usually found adjacent to the coast although case, the central dunefield in Mason Bay, extends
active dunes may occur well inland as remnants, almost 3.5 km inland and is about 1.2 km wide
or where the parent material of earlier phases (Fig. 2). The western boundary is also defined
has been disturbed. Distinctive dune morpho- by the edge of the foredune vegetation. Mobile
logies, including foredunes, parabolic dunes sands are transgressing over a mature podo-
or transverse dunes may be discernible, but carp forest (growing on older dunes) along the
not necessarily so. The seaward extent of an southern boundary. The northern boundary is
active dunefield coincides with the toe of the Duck Creek and an area of broadleaf (Griselinia
vegetated foredune, or seaward limit of incipi- littoralis) forest, which separates the central and
ent foredunes. northern dunefields in Mason Bay. The northern
and southern dunefields are transgressing over
an older dunefield of long-walled parabolic
Methodology
dunes, which is undated, but probably of
Mapping active dunes Late-Holocene age. The central dunefield
Classification of active dunes was, in general, contains a sequence of landforms – a massive
straightforward. Active dunes were identified foredune complex (associated with marram
on the basis of (1) vegetation cover (usually grass), long-walled parabolic dunes, stonefields
sparse); (2) vegetation type (with a predomin- and an extensive sand sheet with a sparse
ance of primary sand colonizing species); (3) vegetation cover. Wetlands, dominated by
Figure 1 Active dunes on the Karikari Peninsula, Northland. The photograph is taken from Puheke (see Fig. 3)
looking east towards Cape Karikari. The active dunes comprise a range of dune forms, including foredunes, deflation
surfaces and transverse dunes. It extends between the toe of the vegetated foredune and the Waimango wetland
and stabilized dunes of Late-Holocene or last interglacial age. Soil development is poor or non-existent and vegetation
cover is sparse.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
110 M. J. Hilton
Figure 2 Active dunes at Mason Bay (central dunefield), located on the west coast of Stewart Island. The photo-
graph is taken looking towards Big Sandhill from above the foredune. The dunefield comprises a mosaic of dune
forms and environments, including a marram-dominated foredune complex, long-walled parabolic dunes, deflation
surfaces (stonefields), sand sheets, interdune wetlands and rock outcrops.
Leptocarpus similis, have recently developed (Fig. 3). In most cases the maps of active
in the deflation zones of parabolic dunes. dunes were derived from multiple runs of aer-
Given the size of the wetlands relative to the ial photography spanning a number of years;
area of the central dunes (272 ha) – the largest hence the maps represent a decade (1950s,
is 0.5 ha – they were included in the estimate 1970s, 1980s and 1990s).
of ‘active dunes’. These two cases were typical The most recent maps produced in the present
of most dune systems. The extent of active study (1990s) were derived from aerial photo-
dunes was usually clearly delineated, because graphs taken for regional councils. Most regional
of the contrast between the relatively sparse councils have obtained low-level vertical, colour,
vegetation of the active dunes and the estab- aerial photography from about 1993. The
lished shrub, or forest cover, of the stable Manawatu-Wanganui Regional Council, for
dunes. example, obtained aerial photographs of their
Four sets of maps have been generated coast between May 1995 and May 1997. In
for each region at a scale of 1 : 63 360 (1950s, addition, the Department of Conservation holds
1970s, 1980s) or 1 : 50 000 (1990s). The maps were more recent aerial photographs in some regions.
mainly compiled from aerial photographs. Comprehensive aerial photography was not
A range of other sources were also utilized, available for five regions/unitary authorities
including topographic maps, soil maps, the (Auckland, Wellington, Nelson, Marlborough,
Coastal Resource Inventory (CRI), the Sand West Coast), and estimates of the area of active
Dune and Beach Vegetation Inventory (Johnson dune in these regions/districts were partly derived
1992; Partridge 1992), PNAP survey reports, from published and unpublished sources,
local government planning maps, published including maps in planning reports and academic
reports, scientific papers, unpublished theses, theses. The aerial photographic interpretations
soil and geological maps, and regional as well were confirmed, in many cases, during field-
as specialist local inventories. Most early work in dune systems in Southland, Otago,
topographic maps represent active dunes very Canterbury, Auckland and Northland in the
precisely, particularly the 1 : 25 000 series period 1995–2005.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 111
Figure 3 Active dunes have been clearly and accurately represented in most topographic map series. This section
of NZMS2 shows the active dunes along a section of Karkikari Bay (shown in Fig. 1).
The boundaries of active dune systems were Draft maps were sent to technical staff in all
identified using the above criteria, then digitized regional and district councils and Department
at the map or photo scale and stored in a of Conservation conservancies for review and
Geographic Information System (ARC-INFO). verification. The initial mapping, based on the
Map data were converted from transverse above sources, identified all but a few active
Mercator projection to New Zealand Map dune systems, and they were subsequently added.
Grid, and regional and subregional maps (3 – 4 Delineation of the boundaries of active dune
per region) generated. ARC-INFO was used systems was difficult in a few cases, particularly
to calculate the area of each active dunefield, where the aerial photographic coverage is
sum the areas of active dunefields in each region poor and the active dune/hinterland vegetation
and produce maps of active dunes in the 1950s, boundary indistinct. Dune system resolution
1970s, 1980s and 1990s for each local govern- varies from a few metres to tens of metres,
ment district/region. These districts and regions depending on the scale and quality of the
are shown in Figure 4, with the location of the aerial photographs used. The regional maps
three subregional maps referred to in the results. have been derived from aerial photographs and
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
112 M. J. Hilton
Figure 4 The boundaries of regional councils and unitary authorities and location of case study dune systems.
maps at a range of scales, but usually between 2005 are presented. The maps represent the
1 : 10 000 and 1 : 63 000. distribution of the dominant foredune species
(Desmoschoenus, Spinifex and Ammophila).
The distribution of marram grass The purpose is to determine the degree to
Marram grass is recognized as invasive and which marram grass has spread in the period
a threat to the ecology of active dunes in 1985–2005 and displaced the above indigenous
New Zealand (Hilton et al. 2005). Maps of the species. Marram grass has been planted in
extent of marram grass and indigenous species many dune systems in backdune environments
around the New Zealand coast in 1985 and (landward of the foredune) in association with
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 113
forestry operations, including many dune sys-
Results
tems in Auckland and Northland so it would
be misleading to incorporate backdune data. National and regional trends in the area of
The 1985 maps are based on site descriptions active dunes
provided in the Sand Dune and Beach Vegeta- The total area of active dunes in New Zea-
tion Inventory of New Zealand (Johnson 1992; land in the early 1900s comprised 128 740 ha
Partridge 1992). The 2005 maps are based on (Cockayne 1911). This is close to the estimate
data derived from information provided by derived from the 1940s/1950s topographic maps
key informants in district and regional councils used in the present study (129 402 ha). Active
and the Department of Conservation. Special- dunes are present in all regions; however, the
ist staff in these organizations were contacted largest systems occur along the west coasts
in 2004/2005, and asked to report the extent of the three main islands. In 1958 active dunes,
and dominance of marram grass in their comprising transgressive dunes in the main,
region/conservancy. This data was augmented occurred along the west coast of the Aupouri
by recent published and unpublished accounts Peninsula (32 100 ha) (Fig. 5) and the Manawatu
of specific dune systems and visits to most coast (12 627 ha) south of the Wanganui River
Northland, Auckland, Southland, Otago and (Fig. 6). The active dunes of the Aupouri Penin-
Stewart Island dune systems in the period sula, which comprise transverse and parabolic
2000–2005. dune forms, extend approximately 80 km along-
The resulting maps indicate the distribution shore and approximately 5 km inland. In both
of the primary foredune species and trends in cases older, stable, dunes form the hinterland.
the spread of marram grass. Three categories The area of active dunes in New Zealand
of marram grass infestation were recognized: declined from 129 402 ha in the 1950s to 38 949
(1) marram grass is the only or dominant ha in the 1990s – a reduction of about 70% over
species; (2) marram grass and indigenous spe- the last 40 years. All regions experienced a
cies occur in more or less equal proportions; decline over this period, although the proportion
and (3) foredunes are dominated by indi- varied (Table 2). The extent of active dunes
genous species. In practice, the ‘dominance’ of declined 7% in the Wellington region, but the
marram grass in the 1980s can only be inferred Manawatu Region lost over 80% (Table 3).
from the site descriptions of Partridge (1992) Regions with the largest area of active dune in
and Johnson (1992) as their inventory con- the 1950s experienced the greatest decline.
tains few quantitative estimates of species Loss of active dunes occurred throughout
cover. the post-World War II period, although at
Figure 5 The area of active dunes on the Aupouri Peninsula has declined as a result of afforestation. Significant
remnants of active dunes survive in a range of reserves and on Måori land near North Cape.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
114 M. J. Hilton
Figure 6 The area of active dunes has also declined in the Manawatu region, as a result of afforestation and
agriculture.
different rates in different regions. Canterbury changed little until the 1970s, when many dune
region experienced a 60% decline between the systems north and south of Dunedin were
1950s and 1970s, largely as a result of stabiliza- stabilized with marram grass and farmed or con-
tion and afforestation of the Pegasus Bay dunes. verted to plantation forestry. The area of active
In contrast, the area of active dune in Otago dune then declined rapidly. Nationally, the rate
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 115
Table 2 Trends in area of active dunes by region
Percentage
Region 1950s (ha) 1970s (ha) 1980s (ha) 1990s (ha) decline (%)
Auckland 15 223.84 8 555.70 4 812.50 NA 68.39
Bay/Plenty 1 692.39 1 363.65 982.26 928.14 45.16
Canterbury 5 207.68 2 075.10 1 928.15 1 846.10 64.55
East Coast 726.49 677.19 693.41 673.03 7.36
Hawkes Bay 2 790.38 2 614.47 1 720.39 1 441.70 48.33
Marlborough 379.46 379.46 247.37 NA 34.80
Nelson NA 512.39 NA NA NA
Northland 64 199.43 38 348.38 24 728.13 15 180.92 76.35
Manawatu 16 627.08 8 600.51 2 636.41 2 359.93 81.31
Otago 1 775.01 1 754.35 1 075.52 1 039.27 41.44
Southland 4 854.55 4 295.54 3 894.75 3 343.02 31.33
Taranaki 1 471.46 1 447.39 419.80 433.45 70.54
Tasman 3 007.05 2 577.75 750.44 654.81 78.22
Waikato 7 168.14 4 600.75 2 155.92 2 022.07 71.79
Wellington 984.89 982.64 914.086 NA 7.2
West Coast 2 781.88 2 435.99 2 399.40 NA 13.75
Total 129 402 81 221 50 012 38 949 69.90
Table 3 Proportion of active dunes in each the national trend. Southland, incorporating
region (1950) (ranked by area) and percentage Fiordland and Stewart Island, lost just 31%
lost (1950s−1990s) of active dunes and the West Coast only 14%.
Percentage of The area of active dunes on Stewart Island
national total Percentage lost (1 258 ha) has not changed significantly since
Region (1950s) (%) 1950s–1990s (%) the first available maps and aerial photographs,
although marram grass has invaded large areas
Northland 49.6 76.3
of the Doughboy Bay and Mason Bay dunes.
Manawatu 12.8 81.3
Stewart Island and Fiordland comprise just
Auckland 11.8 68.4
3.8% and 2.1% of the New Zealand post-
Bay of Plenty 1.3 45.2
War area of active dunes, respectively. How-
Waikato 5.5 71.8
ever, they contain a disproportionately large
Canterbury 4.0 64.5
Southland 3.8 31.3
number of sites identified by Johnson (1992)
Tasman 2.3 78.2 and Partridge (1992) as dune systems of national
Hawkes Bay 2.2 48.3 conservation significance – 23 of the 53
West Coast 2.1 13.7 national priority sites occur in Southland, 10
Otago 1.4 41.4 of these in Fiordland, and 9 on Stewart Island.
Taranaki 1.1 70.5 A further 12 such sites are in Northland,
Wellington 0.8 7.2 although that region lost approximately 76%
East Coast 0.6 7.4 of its active dunes. This pattern is the result
Nelson 0.4 NA of two circumstances. First, marram grass has
Marlborough 0.3 34.8 been in the remote Fiordland and Stewart
Island dune systems for only a few decades;
populations of marram grass on these coasts
of loss of active dune has slowed since the result from long-distance dispersal, rather
1990s, probably because most of the remaining than deliberate and concerted introductions.
active dunefields are in the conservation estate. Second, marram grass does not, in general,
The retention of relatively large areas of displace Desmoschoenus or Spinifex from the
active dunes in some regions is notable, given foredune in the north of the North Island.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
116 M. J. Hilton
Causes of the decline in area of active elvey 1999) following the general planting of
dunes marram grass (Fig. 7). The decline in area of
The decline in area of New Zealand’s active active dunes on the Aupouri Peninsula, for
dunes is largely the consequence of forestry example, following World War II (from 32 100 ha
and agriculture. Urbanization, infrastructure to 8 735 ha), was a consequence of afforestation.
development, sand mining, waste disposal Marram grass is able to disperse and invade
and military activities, have been of secondary active dunes very rapidly, as illustrated by the
importance, particularly near major urban cen- case of Mason Bay, on the west coast of Stew-
tres. Marram grass has been used to stabilize art Island. Farmers at the southern end of
active dunes before their conversion for these Mason Bay introduced the grass in the 1930s
purposes. Marram grass was first recorded (Hilton et al. 2005). In 1958 a few small areas
growing in New Zealand in 1873 at Miramar, of marram grass were present north of Martin’s
near Wellington (Buchanan 1873). By 1900 Creek (8.8 ha), approximately 9 km north of
large quantities were being imported from the liberation site. By 2000, the area of active
Australia, primarily to stabilize active dunes dunes containing marram grass had increased
near major cities. The distribution of marram to 270 ha, which then equalled approximately
grass increased significantly when the (former) 47% of the active dunes north of Martin’s
Lands Department began planting marram Creek (Fig. 8). The rate of invasion during this
following the publication of Cockayne’s 1911 period was exponential (Jul 1998).
report. Cockayne advocated the stabilization Many of the dune systems mapped as ‘active
of sand dunes, with the intention of reducing dune systems’ in the present study are likely to
sand dune migration and establishing produc- have low conservation values where marram
tion forestry. The (former) New Zealand grass has been present for more than 25 years.
Forest Service subsequently developed tech- Such is the case at Mason Bay. Duncan (2001)
nology to stabilize and fertilize active dunes found that marram grass has displaced indigenous
and establish Pinus radiata plantations. Large species from the foredune environment and
areas of formerly active dunes in Northland, caused the development of a new foredune.
Auckland, Manawatu and Waikato were affor- The Department of Conservation has embarked
ested during the 1960s, 1970s and 1980s (McK- on a programme of marram grass eradication
Figure 7 Men planting marram grass on Aupouri Peninsula, Northland, 1910. (A. Northwood, by permission, Alex-
ander Turnbull Library)
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 117
Figure 8 The area of active dunes and active dunes containing marram grass in the central and northern dunes,
Mason Bay. The maps indicate the history of rapid marram grass invasion between 1958 and 2000.
at Mason Bay, which is located in Rakiura tially free of marram grass. Kaitorete Spit,
National Park. south of Christchurch, contains marram grass;
Marram grass is now widespread in New however, the active dunes are still relatively
Zealand and is the main threat to the remain- intact. Marram grass is now the dominant
ing active dune systems. The Sand Dune and species around most of the North Island, with the
Beach Vegetation Inventory surveyed 332 South exception of the northeast coast between North
Island and Stewart Island beaches and 289 North Cape and East Cape and scattered sections
Island beaches during the 1980s (Johnson 1992; of coast north of Hawkes Bay, Wellington and
Partridge 1992). At the time marram grass was Auckland. Indigenous species have not been
the dominant or only foredune species around displaced between North Cape and East Cape
much of the coast of the South Island between on the northeast coast of the North Island,
Cape Foulwind in the northwest and Puysegur except where marram grass is being actively
Point in the southwest (Fig. 9). Marram grass managed in association with forestry opera-
was also found in Fiordland and on Stewart tions (e.g. Pakiri), or in areas of coastal subdivi-
Island, as well as south Westland. Many of the sion and development. It seems likely that,
infestations were small and indigenous foredune within 20 years, marram grass will dominate
species were still common. By the 1980s the foredunes of nearly all New Zealand dune
marram grass had been widely planted in the systems, except where it is being actively
North Island, particularly in the large active controlled. The potential for marram grass to
dune systems of Northland, Auckland and the invade dune systems on the northeast coast
Manawatu. Indigenous species still dominated of the North Island, between East Cape and
the foredunes north of Hawkes Bay on the east North Cape, as well as the west coast of the
coast. However, by this time marram grass was Northland Region, has yet to be determined.
the dominant species on most of the west coast
of the North Island.
Discussion
The cover of marram grass increased signi-
ficantly in the period 1985–2005. In the South The area estimates of active dunes provided in
Island, only two areas, Fiordland and Stewart the present study accord with the early estimates
Island, still contained dune systems substan- of Cockayne (1911), that is, the national area
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
118 M. J. Hilton
Figure 9 Location of ‘national priority dune systems for conservation’ (circles) recommended by Partridge (1992)
and Johnson (1992) – with additional sites identified by the author (triangles) – and sites where marram grass control
is ongoing (open circles). Marram grass has continued to spread and displace indigenous foredune species since the
mid-1980s. Marram grass is present in backdune environments in northern New Zealand, but does not appear to
displace indigenous foredune species. Marram grass has displaced Desmoschoenus in most dune systems in the
Chatham Islands.
of active dunes in the 1950s was very close to and European settlement, respectively. By the
the area around the beginning of the 1900s. 1940s much of the indigenous lowland vegeta-
Newsome’s (1987) estimate of the area of active tion cover of New Zealand had been cleared
dune system at around 52 000 ha is consistent for pastoral farming and agriculture. Dune
with estimates presented here for the late systems were routinely grazed by livestock
1970s, when the New Zealand Land Resource and other introduced species, including rabbits.
Inventory was published (NWASCO 1975 – Dunes and adjacent beaches were being mined
1979). for building sand, and settlements were expand-
It is unclear to what extent the estimates of ing over dune systems. Exotic plant and animal
Cockayne’s (1911) and the results of the present species had colonized most, if not all, active
study indicate the ‘natural’ (pre-human), pre- dune systems. Finally, intensive planting of active
European (<200 years), or pre-Polynesian (<1000 dunes was occurring along the west coast of the
years) area of active dunes. There have so far North Island for plantation forestry (Sale 1985).
been too few studies of dunefield formation These pressures had a severe impact on the
during the Late Holocene in New Zealand. state of certain dune systems, but at the national
Hesp (2001) suggests phases of transgressive level their impact was minor compared to what
dune development in the Manawatu, from happened in the period 1950–2000. The total
600 to 1000 years BP and 150 years BP to the area of active dunes had declined to 39 000 ha
present day, are associated with Polynesian by the early to mid-1990s. Moreover, that figure
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 119
is likely to overestimate the current area of or only low levels of marram grass infestation.
active dunes. The most recent aerial photography Second, there is no ready basis for assessing
available for the Northland Region, which then the extent to which protected and actively
contained the largest area of active dunes, was managed dune systems are representative
flown in September 1993. Conversion of active of regional or national diversity. As yet no
dunes in this and other regions has slowed, but attempt has been made to develop a typology
continued, since that time. of New Zealand dune systems, based on dune
The current study documents the decline form and geomorphic process, despite the
in area of active dunes, not the change in the evidence for distinctive regional variations in
area and diversity of particular dune habitats, the physical and, to a lesser degree, botanical
their ecological integrity, or the conservation character of active dune systems. Third, there
status of dune communities or individual species. have been comparatively few and very localized
Many of the active dunes mapped in the current attempts to map and interpret dune environ-
study are degraded following marram grass ments and associated plant communities. New
invasion and dune stabilization. These dunes Zealand has a poor knowledge of the bio-
are geomorphically inert, although they are diversity of specific dune systems, including
morphologically intact. The impact of marram the threat posed to dune systems by invasive
grass has been particularly severe on the east plants. Emphasis has been placed on marram
and south coasts of the South Island and the west grass in the current paper, in large part because
coast of the North Island. The area of active it is the major threat to active dunes through-
dunes in the Otago Region, for example, has out most of New Zealand. The active dunes
declined by approximately 41% since the 1950s, of Northland are vulnerable to other exotic
a moderate loss by national standards. How- species, including Acacia longifolia.
ever, only two Otago beaches, Tahakopa Bay Existing dune inventories, namely the Sand
and the Tokomairiro River Mouth, retain Dune and Beach Vegetation Inventory of New
significant areas of the indigenous flora usually Zealand and occasional Protected Natural Areas
associated with active dunes. Marram grass Program survey reports, have provided an
comprises over 90% of the vegetation cover in overview of the decline in natural character of
both these dune systems. dune systems and identified active dune systems
Many remaining active dune systems – the of exceptional conservation value. Further
Manawatu dune system, for example (Fig. 6), work is urgently required to (1) develop a bio-
are mere remnants of previously much larger physical classification of active dune habitats
areas. Many of those fragments are of national comparable with that developed for European
significance, despite their reduced area, in dunes; (2) develop inventories of particular dune
that they contain threatened or regionally habitats (e.g. interdune wetlands) and threat-
threatened species. There are now relatively ened or localized dune species; and (3) map
few dune systems with the sequence of semi- and monitor the distribution of key indigenous
vegetated, unstable dunes to stable, vegetated and exotic species. Given the rate at which the
dunes described by Cockayne (1958). At many area and natural character of New Zealand’s
sites the backdunes have been converted to active dunes have declined over the last 40 years,
forestry or agriculture and the foredune is the opportunity for conserving the remaining
dominated by marram grass. active dunes may not last beyond the next
decade.
Conclusions
Acknowledgements
The information required to effectively manage
and conserve New Zealand’s remaining active This project was supported by a University of
dune systems is still far from comprehensive. Otago Research Grant and the Department
The location and boundaries of all remaining of Conservation (Research Grant no. 2494).
dune systems of significant conservation value, My sincere thanks go to Professor Peter
both active and forested, have yet to be deter- Holland for his constructive comments on the
mined. In general, these will be sites with no manuscript.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
120 M. J. Hilton
References (Diploma of Wildlife Management Dissertation).
University of Otago, Dunedin.
Buchanan J (1873). List of plants found on Miramar McFadgen B (1989). Late-Holocene depositional
Peninsula. Transactions of the New Zealand episodes in coastal New Zealand. New Zealand
Institute 6, 349 – 52. Journal of Ecology 12, 145 – 9.
Cockayne L (1909). Report on the sand dunes of New McGlone M (1983). Polynesian deforestation of New
Zealand: The geology and botany with their Zealand: A preliminary synthesis. Archaeology
economic bearing. Department of Lands, in Oceania 18, 11–25.
Wellington. McKelvey P (1999). Sand Forests. Canterbury
Cockayne L (1911). Report on the dune-areas of New University Press, Christchurch.
Zealand: Their geology, botany and reclamation. Muckersie C, Shepherd MJ (1995). Dune phases as
Department of Lands, Wellington. time-transgressive phenomena, Manawatu, New
Cockayne L (1958). The Vegetation of New Zealand. Zealand. Quaternary International 26, 61–7.
Engelmann, Dresden. Newsome PFJ (1987). The vegetative cover of New
Cowie JD (1963). Dune-building phases in the Zealand. Water and Soil Miscellaneous Pub-
Manawatu District, New Zealand. New Zealand lication No. 112, Water and Soil Directorate,
Journal of Geology and Geophysics 6 , 268 – Ministry of Works and Development, Wellington.
80. NWASCO (1975 –1979). New Zealand Land Resource
Doody P (1991). Sand dune inventory of Europe. Joint Inventory worksheets 1: 63 360. National Water
Nature Conservation Committee & European and Soil Conservation Organisation, Wellington.
Centre for Nature Conservation, Tilburg. Olson JS, van der Maarel E (1989). Coastal dunes in
Doody P (2001). Coastal Conservation and Management. Europe – A global view. In: Van der Meulen
Kluwer Academic Publishers, London. F, Jungerius PD, Visser JH, eds. Perspectives in
Duncan M (2001). The impact of Ammophila arenaria Coastal Dune Management. SPB Academic
(Marram grass) on dune communities at Mason Publishing, The Hague, pp. 3 – 32.
Bay, Stewart Island, New Zealand (Master of Partridge T (1992). The Sand Dune and Beach
Science Thesis). University of Otago, Dunedin. Vegetation Inventory of New Zealand. I. North
Hesp P (2001). The Manawatu dunefield: environ- Island. Department of Scientific & Industrial
mental change and human impacts. New Zealand Research, Christchurch.
Geographer 57, 33 – 40. Rapson J (1996). Book review of the Sand Dune and
Hilton MJ, Jul A, Duncan M (2005). Processes of Beach Vegetation Inventory of New Zealand. I.
Ammophila arenaria (marram grass) invasion North Island, II. South Island and Stewart Island.
and indigenous species displacement, Stewart New Zealand Journal of Botany 34, 141–2.
Island, New Zealand. Journal of Coastal Research Ravine DA (1992). Foxton ecological district: Survey
21, 175–85. report for the Protected Natural Areas pro-
Hunter GG, Blaschke PM (1986). The New Zealand gramme. Department of Conservation, Wanganui.
Land Resource Inventory vegetation cover Sale EV (1985). Forest on Sand: The Story of
classification. Water and Soil Miscellaneous Aupouri State Forest. New Zealand Forest
Publication No. 101, National Water and Soil Service, Wellington.
Conservation Authority, Wellington. Shepherd M (1987). Holocene alluviation and trans-
Johnson P (1992). The Sand Dune and Beach Vegeta- gressive dune activity in the lower Manawatu
tion Inventory of New Zealand. II. South Island Valley, New Zealand New Zealand Journal of
and Stewart Island. Land Resources Scientific Geology and Geophysics 30, 175 – 87.
Report Number 16, Department of Scientific & Walls G (1998). Simply sand? Ocean beach dunes,
Industrial Research, Christchurch. Hawkes Bay. Conservation Advisory Scientific
Jul A (1998). Marram grass (Ammophila arenaria) Notes No. 213. Department of Conservation,
invasion of Mason Bay, Rakiura (Stewart Island) Wellington.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
Research Article
Original Article
The spread of marram grass
Blackwell Publishing Asia in NZ
The loss of New Zealand’s active dunes and the
spread of marram grass (Ammophila arenaria)
Michael J. Hilton
Department of Geography, University of Otago, PO Box 56, Dunedin, New
Zealand
Abstract: This article examines the decline of New Zealand’s active dunes in relation to
the introduction of marram grass (Ammophila arenaria). The area of active dunes in New
Zealand declined from 129 000 ha in the early 1900s to about 39 000 ha in 2000; a reduction
of 70%. The extent of active dunes has declined since the 1950s in all regions, particularly
in Northland, Auckland and the Manawatu. The loss of active dunes on the west coast of
the North Island resulted primarily from the introduction of marram grass, followed by the
establishment of Pinus radiata plantations and extensive pastoral farming. Between 1985
and 2005 marram grass extended its range to the detriment of the indigenous foredune
flora. Conservation and resource management agencies should urgently identify dune systems
for conservation management and marram grass eradication.
Key words: afforestation, dune conservation, dune inventory, marram grass.
Coastal dunes of Late-Holocene age are a one widespread strandline species (the recently
distinctive element of the New Zealand coast. introduced Cakile edentula) and only three
Active dunes, the focus of the present paper, indigenous, primary sand-colonizing, foredune
are characterized by ongoing eolian sedimenta- species – Desmoschoenus spiralis (a sedge),
tion and a sparse or patchy vegetation cover. Spinifex sericeus (a grass) and Austrofestuca
They are readily distinguished from older and littoralis (a grass). Compared with European
stable dunes, associated with earlier phases of dunes, the active dunes of New Zealand, in-
dune development (Shepherd 1987; Hesp 2001). cluding the contemporary foredunes, are sparsely
These older dunes were, in the main, forested vegetated. In particular, New Zealand dunes
before European settlement and show advanced lack the diversity of grass species of the
podzol development (Cowie 1963). The contem- genera Ammophila, Leymus, and Elymus, which
porary active dunes of New Zealand contain are widespread on the foredunes of Europe
a diversity of dune environments and plant (Doody 1991, 2001).
communities (Johnson 1992), many of which The active dunes of New Zealand represent
are local equivalents of the European dune the most recent phase of dune mobility and
ecosystem complexes identified by Olson and dunefield formation. The genesis of these
van der Maarel (1989). In other respects, the dunefields has been examined, but not resolved.
active dunes of New Zealand are very different Debate has centred on the relative importance
from European dunes. New Zealand has only of human versus climatic disturbance (McGlone
Note about the author: Mike Hilton is a senior lecturer in the Department of Geography, University of Otago. His current
research is concerned with the interpretation, management and restoration of coastal sand systems (particular
dunes); the dispersal and invasion ecology of exotic dune plants; and methods of coastal management in New
Zealand and Southeast Asia. He is an advocate for the conservation and sustainable management of coastal
resources in relation to a range of coastal resources issues, particularly the mining of coastal sand systems and the
use and restoration of active coastal dunelands.
E-mail: mjh@geography.otago.ac.nz
© 2006 The Author
Journal compilation © 2006 The New Zealand Geographical Society Inc. Published by Blackwell Publishers Ltd.
106 M. J. Hilton
1983; McFadgen 1989). Human use of New Previous inventories
Zealand dunes is very recent compared with
the long history of occupation in Europe. The Several government agencies have prepared
Polynesian settlers of New Zealand may well inventories of ‘coastal dunes’ (Table 1). The
have disturbed, de-vegetated and destabilized purpose and methodology of these surveys
particular dune systems. It is clear, however, has not been consistent. The first survey was
that active dunes have been heavily modified undertaken by Leonard Cockayne in the early
since the arrival of Europeans in New Zealand, 1900s. Cockayne (1911) estimated the total area
primarily as a result of the recent, widespread of ‘dune’ in 1911 to be 128 740 ha, of which
and very rapid stabilization of active dune systems 118 900 ha were located in the North Island
with marram grass (Ammophila arenaria). and 9 840 ha in the South Island and Stewart
The majority of active dunes in New Zealand Island. These estimates excluded established
now bear little resemblance to the accounts shrub and forest on stabilized dunes, but
of Leonard Cockayne, one of New Zealand’s included wetlands within active dune systems.
founding botanists. Since Cockayne’s early The same approach has been adopted in the
descriptions of the flora and vegetation com- present study.
munities of dunes (Cockayne 1909, 1911), The New Zealand Land Resource Inventory
marram grass has been planted throughout (NZLRI) mapped the area of active dunes
New Zealand to stabilize active dunes, establish at a scale of 1 : 1 000 000 (Newsome 1987).
foredunes for property protection and prepare For that purpose, sand dune communities were
dunes for afforestation with North American defined as ‘communities dominated by herba-
conifers, particularly Pinus radiata . These ceous plants and low shrubs occurring on recent,
activities, often in conjunction with sand mining, unstable dune sands’ (Newsome 1987, p. 130).
stock grazing, infrastructure development, Stable backdune habitat, typically dominated
urbanization and the introduction of a wide by woody shrub and tree species, was excluded.
range of exotic plant, shrub and tree species Because of the map scale employed, many
(e.g. Acacia sophorae and Ehrharta villosa smaller active dune systems, particularly along
var. maxima), have contributed to the loss of the east coast of New Zealand, were omitted.
the extent and natural character of active dune The maps of active dune systems were derived
systems. Coastal sand dunes may be New from data collected during the 1960s and
Zealand’s most threatened natural habitat, 1970s for the NZLRI (NWASCO 1975–1979).
measured both in terms of the diminished The total area of active dune corresponding to
extent and modification of vegetation of Newsome’s definition was estimated at 52 000 ha.
remaining areas. A further 40 000 ha of backdune, dominated
New Zealand is a signatory to Agenda 21 by exotic lupins (Lupinus arboreus) and shrubs,
and the Biodiversity Convention. Signatories were identified. In addition, approximately
commit to preparing inventories of critical 200 000 ha of ‘former dunes’ were described as
coastal habitats, conserving and restoring being covered in pasture grasses, pine trees,
critical habitats and identifying critical areas. gorse (Ulex europeaus) and other exotic species
In relation to coastal dunes, progress has (Newsome 1987). Of the 305 000 ha of coastal
been mixed and information on the conserva- sand dunes identified in the NZLRI (NWASCO
tion status of active dunes is incomplete. The 1975–1979), less than 10% were considered
present study aims to (1) identify and map close to their ‘original condition’ (Hunter &
active dunes with the intention of producing a Blaschke 1986). The phrase ‘original condition’
comprehensive national inventory; (2) reveal is presumed here to mean their condition
trends in the reduction of active dune systems around 1900, prior to the widespread stabiliza-
in each region; and (3) document the recent tion of active dunes with marram grass.
spread of marram grass since the first national A national programme to survey the diversity
census of dune flora in the early to mid-1980s. of New Zealand’s indigenous flora, landforms
The strengths and weaknesses of existing dune and landscapes was initiated in 1983. The
inventories are discussed and further know- Protected Natural Areas Programme (PNAP)
ledge requirements identified. aimed to record the location and characteristics
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
Table 1 Comparison of past dune inventories with the present study (LRI – Land Resource Inventory; PNAP – Protected Natural Areas Programme;
LENZ – Land Environments of New Zealand)
Area (ha)
Mapping South Island & National
Authority/Study Source (date) Scale North Island Stewart Island total Comments:
Cockayne (1911) C/F unknown 118 900 9840 128 740 Cockayne included active dunes, but excluded shrubland and forest on
stabilized dunes
Land Resource V (1960/70s) 1 : 1 000 000 52 000 Excludes 40 000 ha of ‘backdune’ dominated by exotic shrubs and 200 000 ha
Inventory of former dunes under pasture and plantation forest (Newsome 1987).
The spread of marram grass in NZ
Protected Natural 1983– 1 : 50 000 Involves sub-regional surveys of ‘ecological districts’ with the objective of
Areas Programme identifying RAPs (recommended areas for protection). Incomplete national
cover; most districts containing dunes have not yet been surveyed.
Sand Dune & F (1984–1988) no maps Assessment of conservation value of most dune systems in New Zealand;
Beach Vegetation no mapping or area estimates of coastal dunes. This study identified 53
Journal compilation © The New Zealand Geographical Society 2006.
Inventory national priority sites for conservation (Johnson 1992; Partridge 1992).
Land Cover S (1996–1997) 1 : 50 000 30 100 11 900 42 000 Maps active dunes, but excludes older vegetated dunes.
Database
Land Environments V (1998) 1 : 50 000 National classification of ‘environments’ based on variation in climatic,
of New Zealand landform and soil variables.
This study C/A (1950s) 1 : 25 000– 110 884 18 518 129 402 Maps active dunes. Excludes older, forested dunes; also maps active dunes
1 : 63 660 in 1950s, 1960s and 1980s by region.
C/A (1990s) 1 : 50 000 28 763 10 041 38 804
Source of information: V – soil and other environmental data; C – cadastral and land cover maps; F – field survey; A – vertical aerial photography; S – satellite data.
Existing classifications are generally based on physical geography and use manual classification techniques to draw visible ecological boundaries onto maps. Newer classifications
© 2006 The Author
such as LENZ are numerically based using computer programs to sort climate, landform and soils data to group areas containing ecosystems of similar type. Both approaches
provide a wealth of information for conservation and resource management.
The numerically based approach means there are no geographic constraints and the classification can be used at varying scales, i.e. environments (containing similar ecosystems)
can vary markedly in size and can be identified even where they are distributed across the landscape in a number of pieces.
107
108 M. J. Hilton
of indigenous plant communities in the know- or animal modification of the site, and the degree
ledge that many remaining communities of high of invasion by weed species. Some large dune
conservation value were poorly represented systems were not visited (e.g. Kokota Spit in
in the existing system of national parks and Northland) and some dune systems, subsequently
reserves. Recommended Areas for Protection identified as having high conservation values
(RAPs) were assessed and identified on the (e.g. Tokomairiro River in Otago), were over-
basis of representativeness, diversity, special looked (Rapson 1996).
features, naturalness, viability, size and shape, The Inventory identified 23 sites in the North
and buffering from external stresses. The bound- Island, 21 in the South Island, and 9 on Stewart
aries of individual RAPs were delineated on Island as ‘national priority dune systems for
topographic maps at a scale of 1 : 50 000. conservation’. All had scores of 15 or more.
The PNAP has provided valuable information Some regions have no sites of national signi-
on the botanical characteristics and conserva- ficance (e.g. Otago). The majority of recognized
tion status of some dune systems, including sites are in the far north and far south of New
those in the Manawatu region (Ravine 1992). Zealand.
Several RAPs contain active or stabilized dunes. As yet, there is no comprehensive database
Most are fragments of much larger, modified of the extent and conservation values of
dunefields. Relatively few ecological districts New Zealand’s active dune systems. The Sand
have been surveyed since the PNAP began in Dune and Beach Vegetation Inventory, despite
1983. There are 179 ecological districts: 104 its methodological limitations, has provided
in the South Island and offshore islands, and the only consistent, nation-wide, assessment of
75 in the North Island. Of these, 123 contain the conservation status of dune systems. It did
stretches of coast. To date, only 33 PNAP not, however, map the boundaries or extent
reports have been published and only 25 of the of individual dune systems, active or forested.
123 coastal ecological districts have been The PNAP surveys provide more detailed
surveyed (18 in the North Island, 7 in the site information, show dune systems of high
South Island). Ecological districts known to conservation value and recommend sites for
contain dune systems of national significance protection, but to date very few dune systems
have yet to be surveyed. Two clusters of dunes have been surveyed. The current character,
of national significance occur in the far north and change in character, of a handful of dune
of the North Island and far south of the South systems have been described (Walls 1998), but
Island (identified by Johnson 1992 and Partridge trends in the state of New Zealand’s dune
1992). The PNAP has not provided the rapid systems have not been documented. There-
synoptic survey of New Zealand’s vegetation fore, the current study identifies trends in the
cover envisaged in the early 1980s. decline in area of active dunes at regional and
The Sand Dune and Beach Vegetation Inven- national scales.
tory (henceforth ‘the Inventory’) conducted by
the Botany Division of the former Department The nature of ‘active dunes’
of Scientific and Industrial Research (Johnson This study is concerned with ‘active dunes’.
1992; Partridge 1992), provided the first national The contemporary ecology of these dunes is
survey and systematic conservation assessment primarily determined by eolian sedimentation.
of coastal dunes in New Zealand (both active Plant diversity is low and mainly restricted
and forested). The aim of that survey was to to species adapted to, or tolerant of, erosion,
identify dune systems with high natural values burial, extreme solar radiation, desiccation and
for conservation purposes (with an emphasis salt spray. Desmoschoenus spiralis (pingao),
on botanical values). Fieldwork was carried out Austrofestuca littoralis (sand tussock), Spinifex
between 1984 and 1988 by regional botanists sericeus, Ammophila arenaria (marram grass)
of the Department of Scientific and Industrial and Euphorbia glauca (shore spurge) are
Research. Individual dune systems were scored commonly associated with active dunes. They
0 to 20, based on the diversity of plant communities tend to form relatively dense stands across
and landforms, the number or proportion of foredunes, but cover may be sparse or patchy
native sand dune species, the degree of human within larger dune systems associated with
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 109
transgressive dunes. A collection of other proximity to the foreshore; and (4) morpho-
species, including the sand daphne (Pimelea logical evidence of transgressive dune forma-
spp.), sand coprosma (Coprosma acerosa), and tion and migration. Two cases, Karikari Beach
Isolepis nodosus, are commonly associated with in the Northland Region, and Mason Bay
active dunes. on Stewart Island (Southland), illustrate how
Active dunes are usually associated with the these criteria were applied. The active dunefield
most recent phase of dunefield development east of Puheke, a conspicuous volcanic cone,
around the New Zealand coast – equivalent to comprises a foredune (dominated by Desmo-
the Hoatan (McFadgen 1989) or the Waitarere schoenus), backed by a sparsely vegetated
phase (Cowie 1963; Muckersie & Shepherd 1995). dunefield that contains multiple transverse
They are readily distinguished from earlier dunes (Fig. 1). The seaward edge of the dune
phases of dune development through soil system is defined by the toe of the vegetated
development, in that they show virtually no foredune. The landward boundary is defined
modification of the grey, quartzo-feldspathic by a wetland. Older dunes, probably of last
sand parent material. This most recent phase interglacial age, form the hinterland. The second
is usually found adjacent to the coast although case, the central dunefield in Mason Bay, extends
active dunes may occur well inland as remnants, almost 3.5 km inland and is about 1.2 km wide
or where the parent material of earlier phases (Fig. 2). The western boundary is also defined
has been disturbed. Distinctive dune morpho- by the edge of the foredune vegetation. Mobile
logies, including foredunes, parabolic dunes sands are transgressing over a mature podo-
or transverse dunes may be discernible, but carp forest (growing on older dunes) along the
not necessarily so. The seaward extent of an southern boundary. The northern boundary is
active dunefield coincides with the toe of the Duck Creek and an area of broadleaf (Griselinia
vegetated foredune, or seaward limit of incipi- littoralis) forest, which separates the central and
ent foredunes. northern dunefields in Mason Bay. The northern
and southern dunefields are transgressing over
an older dunefield of long-walled parabolic
Methodology
dunes, which is undated, but probably of
Mapping active dunes Late-Holocene age. The central dunefield
Classification of active dunes was, in general, contains a sequence of landforms – a massive
straightforward. Active dunes were identified foredune complex (associated with marram
on the basis of (1) vegetation cover (usually grass), long-walled parabolic dunes, stonefields
sparse); (2) vegetation type (with a predomin- and an extensive sand sheet with a sparse
ance of primary sand colonizing species); (3) vegetation cover. Wetlands, dominated by
Figure 1 Active dunes on the Karikari Peninsula, Northland. The photograph is taken from Puheke (see Fig. 3)
looking east towards Cape Karikari. The active dunes comprise a range of dune forms, including foredunes, deflation
surfaces and transverse dunes. It extends between the toe of the vegetated foredune and the Waimango wetland
and stabilized dunes of Late-Holocene or last interglacial age. Soil development is poor or non-existent and vegetation
cover is sparse.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
110 M. J. Hilton
Figure 2 Active dunes at Mason Bay (central dunefield), located on the west coast of Stewart Island. The photo-
graph is taken looking towards Big Sandhill from above the foredune. The dunefield comprises a mosaic of dune
forms and environments, including a marram-dominated foredune complex, long-walled parabolic dunes, deflation
surfaces (stonefields), sand sheets, interdune wetlands and rock outcrops.
Leptocarpus similis, have recently developed (Fig. 3). In most cases the maps of active
in the deflation zones of parabolic dunes. dunes were derived from multiple runs of aer-
Given the size of the wetlands relative to the ial photography spanning a number of years;
area of the central dunes (272 ha) – the largest hence the maps represent a decade (1950s,
is 0.5 ha – they were included in the estimate 1970s, 1980s and 1990s).
of ‘active dunes’. These two cases were typical The most recent maps produced in the present
of most dune systems. The extent of active study (1990s) were derived from aerial photo-
dunes was usually clearly delineated, because graphs taken for regional councils. Most regional
of the contrast between the relatively sparse councils have obtained low-level vertical, colour,
vegetation of the active dunes and the estab- aerial photography from about 1993. The
lished shrub, or forest cover, of the stable Manawatu-Wanganui Regional Council, for
dunes. example, obtained aerial photographs of their
Four sets of maps have been generated coast between May 1995 and May 1997. In
for each region at a scale of 1 : 63 360 (1950s, addition, the Department of Conservation holds
1970s, 1980s) or 1 : 50 000 (1990s). The maps were more recent aerial photographs in some regions.
mainly compiled from aerial photographs. Comprehensive aerial photography was not
A range of other sources were also utilized, available for five regions/unitary authorities
including topographic maps, soil maps, the (Auckland, Wellington, Nelson, Marlborough,
Coastal Resource Inventory (CRI), the Sand West Coast), and estimates of the area of active
Dune and Beach Vegetation Inventory (Johnson dune in these regions/districts were partly derived
1992; Partridge 1992), PNAP survey reports, from published and unpublished sources,
local government planning maps, published including maps in planning reports and academic
reports, scientific papers, unpublished theses, theses. The aerial photographic interpretations
soil and geological maps, and regional as well were confirmed, in many cases, during field-
as specialist local inventories. Most early work in dune systems in Southland, Otago,
topographic maps represent active dunes very Canterbury, Auckland and Northland in the
precisely, particularly the 1 : 25 000 series period 1995–2005.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 111
Figure 3 Active dunes have been clearly and accurately represented in most topographic map series. This section
of NZMS2 shows the active dunes along a section of Karkikari Bay (shown in Fig. 1).
The boundaries of active dune systems were Draft maps were sent to technical staff in all
identified using the above criteria, then digitized regional and district councils and Department
at the map or photo scale and stored in a of Conservation conservancies for review and
Geographic Information System (ARC-INFO). verification. The initial mapping, based on the
Map data were converted from transverse above sources, identified all but a few active
Mercator projection to New Zealand Map dune systems, and they were subsequently added.
Grid, and regional and subregional maps (3 – 4 Delineation of the boundaries of active dune
per region) generated. ARC-INFO was used systems was difficult in a few cases, particularly
to calculate the area of each active dunefield, where the aerial photographic coverage is
sum the areas of active dunefields in each region poor and the active dune/hinterland vegetation
and produce maps of active dunes in the 1950s, boundary indistinct. Dune system resolution
1970s, 1980s and 1990s for each local govern- varies from a few metres to tens of metres,
ment district/region. These districts and regions depending on the scale and quality of the
are shown in Figure 4, with the location of the aerial photographs used. The regional maps
three subregional maps referred to in the results. have been derived from aerial photographs and
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
112 M. J. Hilton
Figure 4 The boundaries of regional councils and unitary authorities and location of case study dune systems.
maps at a range of scales, but usually between 2005 are presented. The maps represent the
1 : 10 000 and 1 : 63 000. distribution of the dominant foredune species
(Desmoschoenus, Spinifex and Ammophila).
The distribution of marram grass The purpose is to determine the degree to
Marram grass is recognized as invasive and which marram grass has spread in the period
a threat to the ecology of active dunes in 1985–2005 and displaced the above indigenous
New Zealand (Hilton et al. 2005). Maps of the species. Marram grass has been planted in
extent of marram grass and indigenous species many dune systems in backdune environments
around the New Zealand coast in 1985 and (landward of the foredune) in association with
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 113
forestry operations, including many dune sys-
Results
tems in Auckland and Northland so it would
be misleading to incorporate backdune data. National and regional trends in the area of
The 1985 maps are based on site descriptions active dunes
provided in the Sand Dune and Beach Vegeta- The total area of active dunes in New Zea-
tion Inventory of New Zealand (Johnson 1992; land in the early 1900s comprised 128 740 ha
Partridge 1992). The 2005 maps are based on (Cockayne 1911). This is close to the estimate
data derived from information provided by derived from the 1940s/1950s topographic maps
key informants in district and regional councils used in the present study (129 402 ha). Active
and the Department of Conservation. Special- dunes are present in all regions; however, the
ist staff in these organizations were contacted largest systems occur along the west coasts
in 2004/2005, and asked to report the extent of the three main islands. In 1958 active dunes,
and dominance of marram grass in their comprising transgressive dunes in the main,
region/conservancy. This data was augmented occurred along the west coast of the Aupouri
by recent published and unpublished accounts Peninsula (32 100 ha) (Fig. 5) and the Manawatu
of specific dune systems and visits to most coast (12 627 ha) south of the Wanganui River
Northland, Auckland, Southland, Otago and (Fig. 6). The active dunes of the Aupouri Penin-
Stewart Island dune systems in the period sula, which comprise transverse and parabolic
2000–2005. dune forms, extend approximately 80 km along-
The resulting maps indicate the distribution shore and approximately 5 km inland. In both
of the primary foredune species and trends in cases older, stable, dunes form the hinterland.
the spread of marram grass. Three categories The area of active dunes in New Zealand
of marram grass infestation were recognized: declined from 129 402 ha in the 1950s to 38 949
(1) marram grass is the only or dominant ha in the 1990s – a reduction of about 70% over
species; (2) marram grass and indigenous spe- the last 40 years. All regions experienced a
cies occur in more or less equal proportions; decline over this period, although the proportion
and (3) foredunes are dominated by indi- varied (Table 2). The extent of active dunes
genous species. In practice, the ‘dominance’ of declined 7% in the Wellington region, but the
marram grass in the 1980s can only be inferred Manawatu Region lost over 80% (Table 3).
from the site descriptions of Partridge (1992) Regions with the largest area of active dune in
and Johnson (1992) as their inventory con- the 1950s experienced the greatest decline.
tains few quantitative estimates of species Loss of active dunes occurred throughout
cover. the post-World War II period, although at
Figure 5 The area of active dunes on the Aupouri Peninsula has declined as a result of afforestation. Significant
remnants of active dunes survive in a range of reserves and on Måori land near North Cape.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
114 M. J. Hilton
Figure 6 The area of active dunes has also declined in the Manawatu region, as a result of afforestation and
agriculture.
different rates in different regions. Canterbury changed little until the 1970s, when many dune
region experienced a 60% decline between the systems north and south of Dunedin were
1950s and 1970s, largely as a result of stabiliza- stabilized with marram grass and farmed or con-
tion and afforestation of the Pegasus Bay dunes. verted to plantation forestry. The area of active
In contrast, the area of active dune in Otago dune then declined rapidly. Nationally, the rate
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 115
Table 2 Trends in area of active dunes by region
Percentage
Region 1950s (ha) 1970s (ha) 1980s (ha) 1990s (ha) decline (%)
Auckland 15 223.84 8 555.70 4 812.50 NA 68.39
Bay/Plenty 1 692.39 1 363.65 982.26 928.14 45.16
Canterbury 5 207.68 2 075.10 1 928.15 1 846.10 64.55
East Coast 726.49 677.19 693.41 673.03 7.36
Hawkes Bay 2 790.38 2 614.47 1 720.39 1 441.70 48.33
Marlborough 379.46 379.46 247.37 NA 34.80
Nelson NA 512.39 NA NA NA
Northland 64 199.43 38 348.38 24 728.13 15 180.92 76.35
Manawatu 16 627.08 8 600.51 2 636.41 2 359.93 81.31
Otago 1 775.01 1 754.35 1 075.52 1 039.27 41.44
Southland 4 854.55 4 295.54 3 894.75 3 343.02 31.33
Taranaki 1 471.46 1 447.39 419.80 433.45 70.54
Tasman 3 007.05 2 577.75 750.44 654.81 78.22
Waikato 7 168.14 4 600.75 2 155.92 2 022.07 71.79
Wellington 984.89 982.64 914.086 NA 7.2
West Coast 2 781.88 2 435.99 2 399.40 NA 13.75
Total 129 402 81 221 50 012 38 949 69.90
Table 3 Proportion of active dunes in each the national trend. Southland, incorporating
region (1950) (ranked by area) and percentage Fiordland and Stewart Island, lost just 31%
lost (1950s−1990s) of active dunes and the West Coast only 14%.
Percentage of The area of active dunes on Stewart Island
national total Percentage lost (1 258 ha) has not changed significantly since
Region (1950s) (%) 1950s–1990s (%) the first available maps and aerial photographs,
although marram grass has invaded large areas
Northland 49.6 76.3
of the Doughboy Bay and Mason Bay dunes.
Manawatu 12.8 81.3
Stewart Island and Fiordland comprise just
Auckland 11.8 68.4
3.8% and 2.1% of the New Zealand post-
Bay of Plenty 1.3 45.2
War area of active dunes, respectively. How-
Waikato 5.5 71.8
ever, they contain a disproportionately large
Canterbury 4.0 64.5
Southland 3.8 31.3
number of sites identified by Johnson (1992)
Tasman 2.3 78.2 and Partridge (1992) as dune systems of national
Hawkes Bay 2.2 48.3 conservation significance – 23 of the 53
West Coast 2.1 13.7 national priority sites occur in Southland, 10
Otago 1.4 41.4 of these in Fiordland, and 9 on Stewart Island.
Taranaki 1.1 70.5 A further 12 such sites are in Northland,
Wellington 0.8 7.2 although that region lost approximately 76%
East Coast 0.6 7.4 of its active dunes. This pattern is the result
Nelson 0.4 NA of two circumstances. First, marram grass has
Marlborough 0.3 34.8 been in the remote Fiordland and Stewart
Island dune systems for only a few decades;
populations of marram grass on these coasts
of loss of active dune has slowed since the result from long-distance dispersal, rather
1990s, probably because most of the remaining than deliberate and concerted introductions.
active dunefields are in the conservation estate. Second, marram grass does not, in general,
The retention of relatively large areas of displace Desmoschoenus or Spinifex from the
active dunes in some regions is notable, given foredune in the north of the North Island.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
116 M. J. Hilton
Causes of the decline in area of active elvey 1999) following the general planting of
dunes marram grass (Fig. 7). The decline in area of
The decline in area of New Zealand’s active active dunes on the Aupouri Peninsula, for
dunes is largely the consequence of forestry example, following World War II (from 32 100 ha
and agriculture. Urbanization, infrastructure to 8 735 ha), was a consequence of afforestation.
development, sand mining, waste disposal Marram grass is able to disperse and invade
and military activities, have been of secondary active dunes very rapidly, as illustrated by the
importance, particularly near major urban cen- case of Mason Bay, on the west coast of Stew-
tres. Marram grass has been used to stabilize art Island. Farmers at the southern end of
active dunes before their conversion for these Mason Bay introduced the grass in the 1930s
purposes. Marram grass was first recorded (Hilton et al. 2005). In 1958 a few small areas
growing in New Zealand in 1873 at Miramar, of marram grass were present north of Martin’s
near Wellington (Buchanan 1873). By 1900 Creek (8.8 ha), approximately 9 km north of
large quantities were being imported from the liberation site. By 2000, the area of active
Australia, primarily to stabilize active dunes dunes containing marram grass had increased
near major cities. The distribution of marram to 270 ha, which then equalled approximately
grass increased significantly when the (former) 47% of the active dunes north of Martin’s
Lands Department began planting marram Creek (Fig. 8). The rate of invasion during this
following the publication of Cockayne’s 1911 period was exponential (Jul 1998).
report. Cockayne advocated the stabilization Many of the dune systems mapped as ‘active
of sand dunes, with the intention of reducing dune systems’ in the present study are likely to
sand dune migration and establishing produc- have low conservation values where marram
tion forestry. The (former) New Zealand grass has been present for more than 25 years.
Forest Service subsequently developed tech- Such is the case at Mason Bay. Duncan (2001)
nology to stabilize and fertilize active dunes found that marram grass has displaced indigenous
and establish Pinus radiata plantations. Large species from the foredune environment and
areas of formerly active dunes in Northland, caused the development of a new foredune.
Auckland, Manawatu and Waikato were affor- The Department of Conservation has embarked
ested during the 1960s, 1970s and 1980s (McK- on a programme of marram grass eradication
Figure 7 Men planting marram grass on Aupouri Peninsula, Northland, 1910. (A. Northwood, by permission, Alex-
ander Turnbull Library)
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 117
Figure 8 The area of active dunes and active dunes containing marram grass in the central and northern dunes,
Mason Bay. The maps indicate the history of rapid marram grass invasion between 1958 and 2000.
at Mason Bay, which is located in Rakiura tially free of marram grass. Kaitorete Spit,
National Park. south of Christchurch, contains marram grass;
Marram grass is now widespread in New however, the active dunes are still relatively
Zealand and is the main threat to the remain- intact. Marram grass is now the dominant
ing active dune systems. The Sand Dune and species around most of the North Island, with the
Beach Vegetation Inventory surveyed 332 South exception of the northeast coast between North
Island and Stewart Island beaches and 289 North Cape and East Cape and scattered sections
Island beaches during the 1980s (Johnson 1992; of coast north of Hawkes Bay, Wellington and
Partridge 1992). At the time marram grass was Auckland. Indigenous species have not been
the dominant or only foredune species around displaced between North Cape and East Cape
much of the coast of the South Island between on the northeast coast of the North Island,
Cape Foulwind in the northwest and Puysegur except where marram grass is being actively
Point in the southwest (Fig. 9). Marram grass managed in association with forestry opera-
was also found in Fiordland and on Stewart tions (e.g. Pakiri), or in areas of coastal subdivi-
Island, as well as south Westland. Many of the sion and development. It seems likely that,
infestations were small and indigenous foredune within 20 years, marram grass will dominate
species were still common. By the 1980s the foredunes of nearly all New Zealand dune
marram grass had been widely planted in the systems, except where it is being actively
North Island, particularly in the large active controlled. The potential for marram grass to
dune systems of Northland, Auckland and the invade dune systems on the northeast coast
Manawatu. Indigenous species still dominated of the North Island, between East Cape and
the foredunes north of Hawkes Bay on the east North Cape, as well as the west coast of the
coast. However, by this time marram grass was Northland Region, has yet to be determined.
the dominant species on most of the west coast
of the North Island.
Discussion
The cover of marram grass increased signi-
ficantly in the period 1985–2005. In the South The area estimates of active dunes provided in
Island, only two areas, Fiordland and Stewart the present study accord with the early estimates
Island, still contained dune systems substan- of Cockayne (1911), that is, the national area
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
118 M. J. Hilton
Figure 9 Location of ‘national priority dune systems for conservation’ (circles) recommended by Partridge (1992)
and Johnson (1992) – with additional sites identified by the author (triangles) – and sites where marram grass control
is ongoing (open circles). Marram grass has continued to spread and displace indigenous foredune species since the
mid-1980s. Marram grass is present in backdune environments in northern New Zealand, but does not appear to
displace indigenous foredune species. Marram grass has displaced Desmoschoenus in most dune systems in the
Chatham Islands.
of active dunes in the 1950s was very close to and European settlement, respectively. By the
the area around the beginning of the 1900s. 1940s much of the indigenous lowland vegeta-
Newsome’s (1987) estimate of the area of active tion cover of New Zealand had been cleared
dune system at around 52 000 ha is consistent for pastoral farming and agriculture. Dune
with estimates presented here for the late systems were routinely grazed by livestock
1970s, when the New Zealand Land Resource and other introduced species, including rabbits.
Inventory was published (NWASCO 1975 – Dunes and adjacent beaches were being mined
1979). for building sand, and settlements were expand-
It is unclear to what extent the estimates of ing over dune systems. Exotic plant and animal
Cockayne’s (1911) and the results of the present species had colonized most, if not all, active
study indicate the ‘natural’ (pre-human), pre- dune systems. Finally, intensive planting of active
European (<200 years), or pre-Polynesian (<1000 dunes was occurring along the west coast of the
years) area of active dunes. There have so far North Island for plantation forestry (Sale 1985).
been too few studies of dunefield formation These pressures had a severe impact on the
during the Late Holocene in New Zealand. state of certain dune systems, but at the national
Hesp (2001) suggests phases of transgressive level their impact was minor compared to what
dune development in the Manawatu, from happened in the period 1950–2000. The total
600 to 1000 years BP and 150 years BP to the area of active dunes had declined to 39 000 ha
present day, are associated with Polynesian by the early to mid-1990s. Moreover, that figure
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
The spread of marram grass in NZ 119
is likely to overestimate the current area of or only low levels of marram grass infestation.
active dunes. The most recent aerial photography Second, there is no ready basis for assessing
available for the Northland Region, which then the extent to which protected and actively
contained the largest area of active dunes, was managed dune systems are representative
flown in September 1993. Conversion of active of regional or national diversity. As yet no
dunes in this and other regions has slowed, but attempt has been made to develop a typology
continued, since that time. of New Zealand dune systems, based on dune
The current study documents the decline form and geomorphic process, despite the
in area of active dunes, not the change in the evidence for distinctive regional variations in
area and diversity of particular dune habitats, the physical and, to a lesser degree, botanical
their ecological integrity, or the conservation character of active dune systems. Third, there
status of dune communities or individual species. have been comparatively few and very localized
Many of the active dunes mapped in the current attempts to map and interpret dune environ-
study are degraded following marram grass ments and associated plant communities. New
invasion and dune stabilization. These dunes Zealand has a poor knowledge of the bio-
are geomorphically inert, although they are diversity of specific dune systems, including
morphologically intact. The impact of marram the threat posed to dune systems by invasive
grass has been particularly severe on the east plants. Emphasis has been placed on marram
and south coasts of the South Island and the west grass in the current paper, in large part because
coast of the North Island. The area of active it is the major threat to active dunes through-
dunes in the Otago Region, for example, has out most of New Zealand. The active dunes
declined by approximately 41% since the 1950s, of Northland are vulnerable to other exotic
a moderate loss by national standards. How- species, including Acacia longifolia.
ever, only two Otago beaches, Tahakopa Bay Existing dune inventories, namely the Sand
and the Tokomairiro River Mouth, retain Dune and Beach Vegetation Inventory of New
significant areas of the indigenous flora usually Zealand and occasional Protected Natural Areas
associated with active dunes. Marram grass Program survey reports, have provided an
comprises over 90% of the vegetation cover in overview of the decline in natural character of
both these dune systems. dune systems and identified active dune systems
Many remaining active dune systems – the of exceptional conservation value. Further
Manawatu dune system, for example (Fig. 6), work is urgently required to (1) develop a bio-
are mere remnants of previously much larger physical classification of active dune habitats
areas. Many of those fragments are of national comparable with that developed for European
significance, despite their reduced area, in dunes; (2) develop inventories of particular dune
that they contain threatened or regionally habitats (e.g. interdune wetlands) and threat-
threatened species. There are now relatively ened or localized dune species; and (3) map
few dune systems with the sequence of semi- and monitor the distribution of key indigenous
vegetated, unstable dunes to stable, vegetated and exotic species. Given the rate at which the
dunes described by Cockayne (1958). At many area and natural character of New Zealand’s
sites the backdunes have been converted to active dunes have declined over the last 40 years,
forestry or agriculture and the foredune is the opportunity for conserving the remaining
dominated by marram grass. active dunes may not last beyond the next
decade.
Conclusions
Acknowledgements
The information required to effectively manage
and conserve New Zealand’s remaining active This project was supported by a University of
dune systems is still far from comprehensive. Otago Research Grant and the Department
The location and boundaries of all remaining of Conservation (Research Grant no. 2494).
dune systems of significant conservation value, My sincere thanks go to Professor Peter
both active and forested, have yet to be deter- Holland for his constructive comments on the
mined. In general, these will be sites with no manuscript.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.
120 M. J. Hilton
References (Diploma of Wildlife Management Dissertation).
University of Otago, Dunedin.
Buchanan J (1873). List of plants found on Miramar McFadgen B (1989). Late-Holocene depositional
Peninsula. Transactions of the New Zealand episodes in coastal New Zealand. New Zealand
Institute 6, 349 – 52. Journal of Ecology 12, 145 – 9.
Cockayne L (1909). Report on the sand dunes of New McGlone M (1983). Polynesian deforestation of New
Zealand: The geology and botany with their Zealand: A preliminary synthesis. Archaeology
economic bearing. Department of Lands, in Oceania 18, 11–25.
Wellington. McKelvey P (1999). Sand Forests. Canterbury
Cockayne L (1911). Report on the dune-areas of New University Press, Christchurch.
Zealand: Their geology, botany and reclamation. Muckersie C, Shepherd MJ (1995). Dune phases as
Department of Lands, Wellington. time-transgressive phenomena, Manawatu, New
Cockayne L (1958). The Vegetation of New Zealand. Zealand. Quaternary International 26, 61–7.
Engelmann, Dresden. Newsome PFJ (1987). The vegetative cover of New
Cowie JD (1963). Dune-building phases in the Zealand. Water and Soil Miscellaneous Pub-
Manawatu District, New Zealand. New Zealand lication No. 112, Water and Soil Directorate,
Journal of Geology and Geophysics 6 , 268 – Ministry of Works and Development, Wellington.
80. NWASCO (1975 –1979). New Zealand Land Resource
Doody P (1991). Sand dune inventory of Europe. Joint Inventory worksheets 1: 63 360. National Water
Nature Conservation Committee & European and Soil Conservation Organisation, Wellington.
Centre for Nature Conservation, Tilburg. Olson JS, van der Maarel E (1989). Coastal dunes in
Doody P (2001). Coastal Conservation and Management. Europe – A global view. In: Van der Meulen
Kluwer Academic Publishers, London. F, Jungerius PD, Visser JH, eds. Perspectives in
Duncan M (2001). The impact of Ammophila arenaria Coastal Dune Management. SPB Academic
(Marram grass) on dune communities at Mason Publishing, The Hague, pp. 3 – 32.
Bay, Stewart Island, New Zealand (Master of Partridge T (1992). The Sand Dune and Beach
Science Thesis). University of Otago, Dunedin. Vegetation Inventory of New Zealand. I. North
Hesp P (2001). The Manawatu dunefield: environ- Island. Department of Scientific & Industrial
mental change and human impacts. New Zealand Research, Christchurch.
Geographer 57, 33 – 40. Rapson J (1996). Book review of the Sand Dune and
Hilton MJ, Jul A, Duncan M (2005). Processes of Beach Vegetation Inventory of New Zealand. I.
Ammophila arenaria (marram grass) invasion North Island, II. South Island and Stewart Island.
and indigenous species displacement, Stewart New Zealand Journal of Botany 34, 141–2.
Island, New Zealand. Journal of Coastal Research Ravine DA (1992). Foxton ecological district: Survey
21, 175–85. report for the Protected Natural Areas pro-
Hunter GG, Blaschke PM (1986). The New Zealand gramme. Department of Conservation, Wanganui.
Land Resource Inventory vegetation cover Sale EV (1985). Forest on Sand: The Story of
classification. Water and Soil Miscellaneous Aupouri State Forest. New Zealand Forest
Publication No. 101, National Water and Soil Service, Wellington.
Conservation Authority, Wellington. Shepherd M (1987). Holocene alluviation and trans-
Johnson P (1992). The Sand Dune and Beach Vegeta- gressive dune activity in the lower Manawatu
tion Inventory of New Zealand. II. South Island Valley, New Zealand New Zealand Journal of
and Stewart Island. Land Resources Scientific Geology and Geophysics 30, 175 – 87.
Report Number 16, Department of Scientific & Walls G (1998). Simply sand? Ocean beach dunes,
Industrial Research, Christchurch. Hawkes Bay. Conservation Advisory Scientific
Jul A (1998). Marram grass (Ammophila arenaria) Notes No. 213. Department of Conservation,
invasion of Mason Bay, Rakiura (Stewart Island) Wellington.
© 2006 The Author
Journal compilation © The New Zealand Geographical Society 2006.