Hilton et al 2005
Journal of Coastal Research 21 1 175–185 West Palm Beach, Florida January 2005
Processes of Ammophila arenaria (Marram Grass)
Invasion and Indigenous Species Displacement,
Stewart Island, New Zealand
Mike Hilton†, Megan Duncan‡, and Anne Jul§
†Department of Geography ‡Department of Conservation §Otago Regional Council
University of Otago Private Bag 68-908 70 Stafford St.
PO Box 56 Newton, Auckland, Dunedin, New Zealand
Dunedin, New Zealand New Zealand
ABSTRACT
HILTON, M.; DUNCAN, M., and JUL, A., 2005. Processes of Ammophila arenaria (marram grass) invasion and
indigenous species displacement, Stewart Island, New Zealand. Journal of Coastal Research, 21(1), 175–185. West
Palm Beach (Florida), ISSN 0749-0208.
The present study (1) describes the rates and patterns of Ammophila arenaria (marram grass) invasion in a large
transgressive dune system (Mason Bay) and on a prograding foredune-ridge barrier (Doughboy Bay), Stewart Island,
New Zealand; (2) examines the impact of Ammophila on dune morphology and indigenous dune biota; and (3) assesses
the significance of geomorphic processes in accounting for the patterns observed. Processes of Ammophila invasion
are interpreted from evidence of landform development and vegetation change; field observations and survey of dune
landforms and dune vegetation; the aerial photographic record and historic accounts of the local botany.
The area dominated by Ammophila in the Mason Bay study area has increased from 1.4 ha in 1958, to 17.8 ha in
1978, to 74.9 ha in 1998; a 5,204 percent increase. Ammophila invasion of active dune systems in the study areas is
clearly associated with dune forming processes—shadow dune development; migration of long-walled parabolic dunes;
stoss face blowout development; and barrier progradation. The primary mechanism of native species displacement
appears to be burial rather than competition for nutrients. Ammophila traps sand and builds dunes at rates that may
exceed the threshold of tolerance of local native species. Desmoschoenus spiralis, the dominant indigenous foredune
species, cannot co-exist with Ammophila in the active dune systems investigated.
ADDITIONAL INDEX WORDS: Desmoschoenus spiralis, foredune complex, parabolic dune migration, Mason Bay,
Doughboy Bay, burial, shadow dune.
INTRODUCTION To date, the processes of Ammophila invasion and indige-
nous plant displacement have only been described in general
Ammophila arenaria (Marram Grass or European Beach-
terms. For example, BUELL, PICKART and STUART (1995,
grass) has been introduced to the active dune systems of
p.1591) attribute the rapid and successful spread of Ammo-
southeast Australia (HESP and THOM, 1990; CULLEN, 1998);
phila through the dunes of North Spit, Humboldt Bay, Cali-
North America (WIEDEMANN and PICKART, 1996); South Af-
fornia, to ‘‘a number of processes and conditions: optimal hab-
rica (HERTLING and LUBKE, 1999a); Chile (HULTEN and
FRIES, 1986; CASTRO, 1988) and New Zealand (JOHNSON, itat, multiple introduction, natural and human disturbance,
1992; PARTRIDGE, 1992). At most sites it has been planted to and proximity to the strand where rhizome fragments are
construct or re-establish foredunes or stabilise transgressive washed ashore by storm surf’’. Increasing our understanding
dune systems. of these processes will assist conservation agencies assess the
Ammophila has belatedly been shown to be highly invasive vulnerability of dune systems to Ammophila invasion and in-
and a threat to the ecology of active dune systems outside its crease the effectiveness of Ammophila eradication operations.
natural range. It may adversely affect indigenous dune flora The impact of Ammophila on the native dune flora and
and the diversity of habitats by stabilizing naturally mobile fauna of New Zealand has been recognised for some time (e.g.
dunes and accelerating vegetation succession (COOPER, 1958; JOHNSON, 1992; DEPARTMENT OF CONSERVATION, 1997).
BUELL, PICKART and STUART, 1995; HERTLING and LUBKE, New Zealand has three indigenous foredune sandbinders,
1999b). The area of bare sand, species diversity and evenness Desmoschoenus spiralis (a sedge), Austrofestuca littoralis (a
(uniformity in abundance and cover of species) has been perennial grass) and Spinifex sericeus (a perennial grass).
shown to correlate negatively with Ammophila along the west Desmoschoenus and Austrofestuca occur throughout New Zea-
coast of the United States (reviewed by PICKART and SAW- land while Spinifex occurs throughout the North Island and
YER, 1998) and in Tasmania (reviewed by CULLEN, 1998). in northern South Island. Ammophila has displaced Desmos-
choenus in many dune systems, but particularly around the
01041 received 3 May 2001; accepted in revision 25 July 2002. south and east coasts of the South Island and the exposed
176 Hilton, Duncan and Jul
western coasts of the three main islands. Ammophila is al- Bay dunefield is dominated by parabolic dune forms. Am-
most the only foredune species on exposed western coasts and mophila was planted near Kilbride at the southern end of
around the south coast of the South Island of New Zealand. Mason Bay in the early 1930s (Figure 1) and east of the
A recent inventory of threatened plant species in New Zea- ‘Homestead’ in the mid 1960s (Figure 2). Ammophila now
land concluded Ammophila was a threat to ten indigenous dominates the foredune environment but is patchy and pre-
dune species, including Desmoschoenus and Austrofestuca sent in very low densities across approximately 60 per cent
(REID, 1998). of the hinterland of the active dunefield between Martin’s
The displacement of indigenous dune plant species in mo- Creek and Duck Creek. Austrofestuca littoralis and Desmos-
bile dune systems appears to be related to geomorphic pro- choenus remain the primary sandbinders landward of the
cesses of dune development. Ammophila-dominated dunes stonefield.
are generally steeper and higher than dunes formed in as-
sociation with the indigenous foredune species, Desmoschoen-
Doughboy Bay
us spiralis and Spinifex sericeus in New Zealand (ESLER,
1970; HOLLAND, 1981). PARTRIDGE (1995) concluded Am- The active dune systems of Doughboy Bay provide oppor-
mophila can either co-exist with Desmoschoenus or displace tunities to study processes of marram grass invasion on a
it, depending on several factors. The species can co-exist in rapidly prograding bay-head barrier. Until recently the series
three situations: (1) on the front of the foredune where mar- of foredune ridges that comprise the seaward half of the
ram grass spread is limited by high salt concentrations; (2) southern barrier had not experienced significant disturbance.
where moisture in the upper sand layers is not limiting; and These are primary dune forms that have developed in con-
(3) where both species are moribund because of low rates of junction with Ammophila. The foredune ridges formed since
sand accumulation. Displacement of Desmoschoenus was the 1960s, probably as a result of post-storm sedimentation.
found to be most severe on stable dunes and where moisture
in the upper layers of sand was limiting. Desmoschoenus dis-
placement was found to be unrelated to the form and geo- METHODS
morphic processes of different kinds of dunes.
The process of Ammophila dispersal and invasion in active
Patterns of Ammophila Invasion
dune systems is not well documented in New Zealand, partly
The distribution and density of Ammophila in the Mason
because most dune systems were completely occupied by Am-
Bay study area were mapped from vertical aerial photo-
mophila by the mid 1900s. Specific geomorphic conditions
graphs flown in 1958, 1978 and 1998. The 1958 photographs
that aid or hinder Ammophila invasion have not been de-
are relatively poor black and white images, but of sufficient
scribed. The present study (1) describes the rates and pat-
quality to allow the identification of areas of Ammophila. The
terns of Ammophila invasion at two sites on the west coast
remainder are high quality colour images, printed at 1:3,000
of Stewart Island, New Zealand; (2) evaluates the importance
and 1:10,000. Resolution of contrasting features is approxi-
of geomorphic processes in accounting for these patterns; and
mately 1 m2. The southern dune system of Doughboy Bay was
(3) interprets the rate and processes of Desmoschoenus dis-
photographed in 1977, 1987 and 1999. The 1977 Doughboy
placement by Ammophila.
image is also black and white. Maps of Ammophila distribu-
STUDY AREAS tion and density derived from the 1998 (Mason Bay) and 1999
(Doughboy Bay) aerials were field checked during 1999.
Mason Bay Maps of the location and density of Ammophila between
The combination of available sediment and prevailing on- Duck Creek and Martin’s Creek were constructed using ARC/
shore winds has led to extensive dune building and trans- INFO software. The 1:3,000 1958 image was digitised by
gressive dunefield development along the western and north- hand. The 1978 and 1998 images were scanned at 1:10,000.
ern coasts of Stewart Island. Prevailing winds are from the This permitted on-screen digitising of these images using the
west (Figure 1). The active dune systems of Stewart Island 1:3,000 images for reference. All Mason Bay maps were or-
encompass a range of primary and secondary dune forms in tho-rectified based on GPS observations of distinctive fea-
a variety of depositional environments. Mason Bay contains tures.
the largest transgressive dune system on Stewart Island and Five density classes of Ammophila were mapped in the Ma-
one of the largest and least modified dune systems in New son Bay study area (0–5, 6–25, 26–50, 51–75 and greater
Zealand. Between Martin’s Creek and Duck Creek, the active than 75 percent). Classes refer to the proportion of ground
dune system extends up to 3 km inland. The foredune com- covered by Ammophila within discrete areas. These areas
plex is a large, continuous feature up to 15 m high and 150 generally correspond with the boundaries of particular land-
m wide. Landward of the foredune a series of active U- forms, for example, the depositional lobes of parabolic dunes.
shaped, long-walled parabolic dunes (after PYE, 1983; HESP Ammophila is considered the ‘dominant’ species where it com-
and THOM, 1990) are transgressing a broad, gently-sloping prises more than 50 percent ground cover within a discrete
stonefield. These dunes are evident in 1958 aerial photo- area. Just one density class is recognised at Doughboy Bay
graphs and formed prior to the arrival of Ammophila in the ( 50 per cent). Ammophila occupied 100 percent of the active
central area of Mason Bay. The hinterland of the dune system dune system seaward of the storm scarp in 1977 at densities
north of Martin’s Creek to the northern end of the Mason exceeding 50 percent ground cover.
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 177
Figure 1. Location of study sites and major Stewart Island active sand dune systems. Wind rose derived from hourly observations, Southeast Cape,
Stewart Island (1992–97).
Landform Development and Vegetation Change dune complex (40); (iii) the deflation zone of the long-walled
parabolic dunes (40); (iv) trailing arms (40); and (v) deposi-
Contemporary vegetation cover in the Mason Bay study tional lobe (25). The average ground cover for each stratum
area was estimated by random sampling (1 m2 quadrat) in was derived by averaging the percent cover of the major spe-
January 2000. The quadrat data was then stratified into five cies observed in each quadrat. Our interpretation of species
classes based on geomorphic characteristics: (i) the stoss face diversity at Doughboy Bay is based on data gathered over a
of foredune complex (40 sites); (ii) the lee slope of the fore- three year period (1999–2001) from ten permanent quadrats
Journal of Coastal Research, Vol. 21, No. 1, 2005
178 Hilton, Duncan and Jul
Mason Bay in 1908, prior to the introduction of Ammophila
(COCKAYNE, 1909). In addition, Ammophila-dominated fore-
dunes in the Mason Bay study area were compared with rem-
nant foredunes in northern Mason Bay. Desmoschoenus is
still the primary sandbinder along small stretches of coast-
line north of Duck Creek and is the dominant sandbinder
across large areas of the hinterland.
RESULTS
Patterns of Ammophila Invasion
Ammophila spread north in Mason Bay following introduc-
tion at Kilbride by farmers in the 1930s. By 1958 Ammophila
had established a small number of dense colonies north of
Martin’s Creek (Figure 2). The total area of dense Ammophi-
la, where Ammophila cover exceeded 50 percent, was around
1.4 ha. At that time almost all Ammophila occurred within
400 m of the toe of the foredune. Further north, beyond the
advancing Ammophila, the foredune comprised a band of low
shadow and/or coppice dunes, approximately 40 m wide. The
indigenous foredune vegetation in 1958 was sparse (compris-
ing, perhaps, 10–30 per cent surface cover). Desmoschoenus,
Austrofestuca and Euphorbia glauca were probably common
at foredune sites and backdune sites experiencing high rates
of sedimentation. Euphorbia glauca was described by COCK-
AYNE (1909) as common and was still relatively abundant in
the 1960s. Pimelea lyallii (Sand Daphne), Coprosma acerosa
(Sand Coprosma), Gentiana saxosa and Raoulia hookeri were
probably also widespread in relatively sheltered backdune
and deflation areas. The long-walled parabolic dunes de-
scribed above lay to landward of this foredune environment
in 1958.
The extent and density of Ammophila increased between
1958 and 1978 (Figure 2). By 1978, Ammophila was present
in dense patches up to 750 m inland. A continuous band of
Ammophila occupied the former foredune environment. The
width and density of this band decreased towards Duck
Creek. A continuous, though topographically irregular fore-
dune was present (Stage III after HESP, 1988). The extent
and density of Ammophila in the study area continued to in-
crease between 1978 and 1998. During this period Ammo-
phila also spread from the 1960s plantings (east of the Home-
stead) about 700 m along the north-eastern margin of the
Figure 2. Ammophila arenaria invasion of Mason Bay study area, 1958–
1998, based on aerial photography interpretation. Only the 1998 pattern dune system (Figure 2).
was field checked. The proportion of the study area dominated by Ammophila
increased from 1.4 ha in 1958 to 17.8 ha in 1978 to 74.9 ha
in 1998. Therefore, the area dominated by Ammophila in-
creased 5,204 per cent in the period 1958–1998. The area
located across the Ammophila-dominated section of the with little or no Ammophila (0–5 percent ground cover) de-
southern dune barrier (which comprises about half of the to- clined 46 per cent over the same period.
tal area of the barrier). These quadrats were established to Available aerial photographs show that Ammophila spread
monitor vegetation change during an Ammophila eradication from the south in Mason Bay (taking approximately 25 years
programme, which commenced in January 1999. to form a continuous foredune between Martins Creek and
Historic vegetation cover (1908–present) and associated Duck Creek). Ammophila invasion of the transgressive dune
landforms were interpreted from aerial photographs, a sur- system in Masons Bay is continuing. In contrast, Ammophila
vey of dune landforms and dune biota (including dead plant became the dominant foredune and backdune species at the
material, especially Desmoschoenus root material); and early Doughboy Bay site over a much shorter period. The area of
botanical accounts. Leonard Cockayne, one of New Zealand’s Ammophila increased rapidly between 1977 and 1999 as the
pioneering botanists, described the dunes and dune flora of barrier prograded by the formation of foredune ridges, from
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 179
Landform Development and Vegetation Change
Since the arrival of Ammophila the foredune in central Ma-
son Bay study area has undergone considerable change.
COCKAYNE (1909) described aspects of the pre-Ammophila
dune landscape and dune flora of Mason Bay. In 1908 the
foredune comprised a line of ‘‘low hummocks’’, 2–3 m high.
The probable position and elevation of these dunes relative
to the modern foredune complex is shown in Figure 4. COCK-
AYNE’s description of the foredune is consistent with the dune
landscape as can be discerned from the 1958 aerial photo-
graphs, soon after Ammophila established north of Martins
Creek. The aerial photographs indicate a hummocky, discon-
tinuous foredune, probably equivalent to the Stage IV fore-
dune of HESP (1988).
The foredune complex developed between 1958 and 1978
as adjacent shadow dunes coalesced and Ammophila estab-
lished a semi-uniform vegetation cover. The foredune is now
a relatively massive landform termed a ‘foredune complex’
(after DOING, 1985). Deposition of sand across the Mason Bay
foredune complex has continued as a result of: (1) formation
of minor blowouts affecting the stoss face of the foredune; and
(2) erosion and transport of sands across the beach during
episodes of strong onshore winds. During the latter, sand is
transported and deposited across the width of the foredune
complex, well inland of the crest. Both processes appear to
contribute to the vertical accretion and lateral growth of the
foredune complex. Between 1978 and 1998 the morphology of
the stoss face of the foredune complex has become more reg-
ular as the density of Ammophila increased and the extent
and frequency of foredune blowouts declined. Narrow blow-
outs of a few metres wide and 20 m or so deep still occur, but
these appear short-lived and are minor compared with the
mass of the foredune complex. Periodic scarping during storm
conditions limits the growth of Ammophila down the lower
slopes of the stoss face of the foredune complex; hence, the
plan configuration of the seaward edge of the foredune com-
plex has become very regular since the establishment of Am-
mophila (Figure 2).
The long-walled parabolic dunes in the central Mason Bay
study area developed before Ammophila invasion—they are
present in the 1958 aerial photographs and were briefly de-
scribed by COCKAYNE (1909). They are active transgressive
dunes. Their depositional lobes have migrated at average
rates of 5.0–7.5 m per year since 1978. The morphology and
Figure 3. Morphology and plant cover of parabolic dune number six (in plant cover of parabolic dune number six is typical of these
1998). The contemporary dimensions of the foredune complex are com-
pared with Cockayne’s (1909) description of the foredune in 1908.
landforms. Four morphologic components are recognisable:
(1) the deflation zone; (2) an area of active erosion incorpo-
rating the ‘throat’; (3) a depositional lobe comprising a com-
plex of coppice and shadow dunes; and (4) shared trailing
about 1.7 ha in 1977 to about 7.0 ha in 1999, a 412 per cent arms (Figure 4). These elements of the parabolic dune have
increase. At the same time the barrier prograded seawards been maintained as the throat and depositional lobe ad-
(45–60 m) and towards the north (140 m) (Figure 3). The vanced inland and the trailing arms lengthened in the period
foredune ridges formed since 1977 have been colonised by 1978 to 1998.
Ammophila and now Ammophila is the only species inhabit- Live Desmoschoenus was observed during 1999 in the de-
ing the stoss face of the modern foredune. There is no evi- flation zones of most of the parabolic dunes in the study area,
dence, exposed rhizome for example, that Desmoschoenus has (although most was moribund); on the level surfaces of the
ever formed a significant cover on the foredune ridges, trailing arms (with Ammophila and a range of native dune
though a handful of isolated plants were observed in 1999. and opportunistic species); and, occasionally, forming isolated
Journal of Coastal Research, Vol. 21, No. 1, 2005
180 Hilton, Duncan and Jul
Figure 4. Plant species associated with the foredune complex (stoss and backdune slopes), trailing arms, deflation zones and depositional lobes. Values
based on random survey of all long-walled parabolic dunes between Duck Creek and Martins Creek.
shadow dunes in the depositional lobes. Aerial photography bly present (Figure 6b), or Ammophila arrived soon after the
(1958) and the widespread occurrence of dead root material storm. Subsequent, post-storm, progradation occurred in
in the eroding faces of the trailing arms and across the throat conjunction with Ammophila, not Desmoschoenus (Figure 6c
area shows that Desmoschoenus was, until recently, the & 6d) as successive foredune ridges established seawards of
dominant sand-binding species on these dunes. These dunes the former storm scarp. Ammophila now forms a dense cover
are now evolving in conjunction with Ammophila, which ac- across the stoss face of the contemporary foredune and a
counts for about half the plant cover across the trailing relatively sparse cover further inland. Desmoschoenus is ab-
arms, less than half the total plant cover in the deflation sent from the foredune ridge section of the barrier. The
zones and over 90 percent of the plant cover in the deposi- storm scarp, therefore, separates the older Desmoschoenus
tional lobes (Figure 5). barrier from the foredune ridge landscape associated with
The pre-Ammophila barrier in Doughboy Bay probably Ammophila. Ammophila and Desmoschoenus occur land-
comprised a relatively simple foredune complex, 100–150 m ward of the scarp, where Gaultheria macrostigma and other
wide (Figure 6a). Some time in the mid 1900s (ca. 1960) an creeping species form a dense mat and where sand move-
exceptional storm event eroded the front of the barrier. At ment is nil. Neither species will survive if stability contin-
this time both Desmoschoenus and Ammophila were proba- ues.
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 181
Figure 5. Recent progradational history of the southern dune system, Doughboy Bay, 1977–1999. Progradation has occurred in conjunction with Am-
mophila arenaria. Desmoschoenus spiralis is now restricted to the remnant barrier landward of the storm scarp.
Figure 6. Interpretation of coastal change and Ammophila invasion, southern dune system, Doughboy Bay, ca. 1900–1999.
Journal of Coastal Research, Vol. 21, No. 1, 2005
182 Hilton, Duncan and Jul
DISCUSSION Creek and Duck Creek. Over the same period, Desmoschoenus
and Austrofestuca were displaced from the foredune and gen-
Ammophila Dispersal and Invasion erally lost from the adjoining parabolic dunes. These pro-
The establishment of Ammophila in the Mason Bay study cesses of foredune development and indigenous foredune spe-
area probably resulted from clonal spread, since reproduction cies displacement are ongoing north of Duck Creek. Austro-
by seed is thought to be exceptional (HUISKES, 1979). Frag- festuca littoralis and Euphorbia glauca, described by COCK-
ments of rhizome are tolerant of seawater rafting for at least AYNE as common in 1908, are now also absent from the
8 tidal cycles (BAYE, 1990) and probably much longer (e.g. foredune and parabolic dune environment in the Mason Bay
HEYLIGERS, 1985). Fragments of rhizome could have entered study area.
the sea in southern Mason Bay, during foredune scarping; Burial is probably the principle mechanism of indigenous
and then been transported alongshore by nearshore currents species displacement. Individual Ammophila plants respond
and into the foredune environment by storm wave swash. to burial by rapid production of elongated stem internodes
This process may have occurred during a single storm event (HUISKES, 1979). Shoots develop from these nodes, which
or phase of storminess, since most of the patches of Ammo- then produce leaves upon reaching the surface. This process
phila in 1958 in the study area appear of similar age (3–5 is repeated if sand supply continues, eventually producing a
years). tussock habit (CHAPMAN, 1964). These tussocks create eddies
Ammophila is most likely to have established in the inter- or vortices in the lee of the plant during phases of onshore
dune hollows of the study area, some distance inland of the sand transport that tend to produce a pyramidal-shaped
mean high tide line, following deposition of rhizomes during shadow dune (HESP, 1981). Ammophila shadow dune devel-
storm conditions. Storm-wave deposition of rhizome propa- opment results in areas of rapid accretion. As individual
gules would have been assisted by the discontinuous foredune dunes develop intervening depressions may experience in-
as it existed in 1958. The contemporary Desmoschoenus fore- tense erosion. Sand deposited in the lee of these dunes, or as
dunes north of Duck Creek are typically low shadow dunes depositional lobes downwind of interdune deflation areas,
2–4 m high, with multiple interdune openings at about the may extend the influence of the first line of Ammophila shad-
level of the spring high tide line. Interdune hollows in the ow dunes well inland (Figure 7a). Desmoschoenus located in
foredune are flooded during storm events and commonly the lee of the initial Ammophila plant and in the new inter-
choked with flotsam and wind-blown material. dune hollows may experience lethal rates of accretion and
Successful Ammophila dispersal may rely on extreme erosion, respectively.
storm events that are able to both erode propagules from the These processes may, over time, establish a massive fore-
face of the foredune and deposit them in suitable locations. dune complex where none existed previously. Subsequent
Ammophila is intolerant of substrate salt concentrations in blowout development across the stoss face may contribute to
excess of 1–1.5 percent (HUISKES, 1979; CHAPMAN, 1964) and the vertical accretion of the foredune complex (Figure 7b),
may have difficulty establishing low on the stoss face of Des- which may also work to maintain Ammophila vigor. All plant
moschoenus foredunes, in circumstances where inundation is species experience stress in the deflation areas of these blow-
common. Ammophila rhizomes blown or washed well inland, outs, however, Ammophila plants on the margins of the de-
beyond the usual reach of storm waves, may stand the best flation basin and across the depositional lobes experience re-
chance of survival. Successful establishment of Ammophila newed growth following accelerated sand deposition. Subse-
may also depend on favourable meteorological conditions fol- quent accretion and recolonisation by Ammophila may rap-
lowing transport and deposition of rhizomes. Ammophila es- idly close the blowout.
tablishment, therefore, may be episodic, even where the sup- Ammophila invasion and concomitant dune development in
ply of rhizome material is constant. the Mason Bay study area appears to result in rates of sand
Two other factors may have affected the potential for Am- accretion (and erosion) that exceed the tolerance of native
mophila to establish on or close to the stoss face of the former dune species. Ammophila is known to accelerate rates of sand
foredune in Mason Bay. Compared with coasts in southeast accretion (CHAPMAN, 1964) and burial is recognised as a
Australia, Europe and elsewhere, the study site lacks her- strong selective force in dune ecology (MAUN, 1998). The abil-
baceous pioneer annuals (the ‘ephemerous tidemark’ and ‘pe- ity of a plant to survive burial includes the plants tolerance
rennial tidemark’ communities of DOING (1985)). Species of darkness and capacity to grow up through sand. Ammo-
such as Cakile maritime may provide suitable conditions for phila is able to tolerate rates of sand accumulation of up to
Ammophila establishment across incipient foredunes but 1 m per year (WILLIS et al., 1959; RANWELL, 1972), although
have only recently established in the study area. Secondly, experimental work indicates instantaneous total burial will
berm development is unusual along Mason Bay beach and kill most specimens (SYKES and WILSON, 1990a). However,
waves frequently cut the toe of the foredune. SYKES and WILSON (1990b) have demonstrated that Ammo-
phila is much more tolerant of darkness (surviving burial for
Indigenous Species Displacement 117 days) than the native foredune species; Austrofestuca (37
days); Euphorbia (50 days); or Desmoschoenus (67 days). Am-
Displacement of Desmoschoenus and Austrofestuca occurred mophila may possess other competitive advantages (tolerance
rapidly following Ammophila invasion in central Mason Bay of drought, for example), but these are probably of minor im-
and Doughboy Bay. Between 1958 and 1978 Ammophila es- portance compared to the ability of this species to initiate and
tablished a single, continuous foredune between Martin’s survive rapid sand accumulation.
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 183
Figure 7. Interpretations of landform development and Ammophila invasion in the study areas.
Displacement of Desmoschoenus across the long-walled the survival threshold of these species. At the same time
parabolic dunes in Mason Bay has also been rapid since the the indigenous species in the depositional lobe and throat
establishment of Ammophila. This process is an inevitable zones are destroyed as the dune advances downwind. The
consequence of dune migration. The throat and depositional trailing arms are relatively stable and hence retain higher
lobes are eroded as the dunes advance. The eroded sand is indigenous species diversity. Secondly, where Desmoschoen-
transported downwind and landward, usually through a us is able to establish in the depositional lobes, it is probably
narrow throat, or multiple throats, to contribute to the pro- displaced by burial.
gressive construction of a new depositional lobe, progres- Finally, Ammophila invasion may be very rapid on
sively landward of the former (Figure 7c). In this fashion, coasts experiencing progradation (Figure 7d). Associated
the depositional lobes are being continually recycled. The loss of native dune species may be equally rapid. Ammo-
destruction of Desmoschoenus is, therefore, a natural pro- phila colonized developing foredunes at Doughboy Bay fol-
cess. The failure of Desmoschoenus to re-establish in the de- lowing a significant storm event more rapidly than Des-
positional lobes and trailing arms is a consequence of com- moschoenus. In this case Ammophila invasion and Desmos-
petition with Ammophila. Ammophila appears better able choenus displacement may be very rapid, of the order of
and more aggressive at colonising the depositional lobes of years. In such dynamic circumstances inter-specific com-
these landforms. Once established, Ammophila may dis- petition for nutrients is unlikely to be a significant dis-
place indigenous species by accelerating accretion beyond placement mechanism.
Journal of Coastal Research, Vol. 21, No. 1, 2005
184 Hilton, Duncan and Jul
CONCLUSIONS interference of dune-beach interaction, Chile. Journal of Coastal
Research Special Issue, 3, 103–107.
In summary, Ammophila invasion and Desmoschoenus dis- CHAPMAN, V.J., 1964. Coastal Vegetation. London: Pergamon Press,
placement in Mason Bay and at Doughboy Bay is associated 245p.
COCKAYNE, L., 1909. Report on a Botanical Survey of Stewart Island.
with at least four processes of landform development (Figure Wellington: Government Printers, 66p.
7): (a) shadow dune and foredune development; (b) blowout COOPER, W.S., 1958. Coastal Sand Dunes of Oregon and Washing-
development; (c) parabolic dune migration; and (d) barrier ton. Memoir 72, Geological Society of America, Boulder, Colora-
progradation. do, 85p.
CULLEN, P., 1998. Ammophila arenaria and Euphorbia paralias: Se-
Ammophila invasion of the active dune systems examined
rious Threats to the Integrity of the South West Tasmanian Coast-
is clearly associated with dune forming processes. Competi- line. Unpublished Report, Hobart: Tasmanian Parks and Wildlife
tion for nutrients is almost certainly important in semi-stable Service, 18p.
situations, but not in dynamic geomorphic situations. In such DEPARTMENT OF CONSERVATION, 1997. Ecology and Management of
circumstances Ammophila traps sand and builds dune at Invasive Weeds. Conservation Sciences Publication No. 7. Welling-
ton: Department of Conservation, 67p.
rates that exceed the threshold of tolerance of all native spe- DOING, H., 1985. Coastal foredune zonation and succession in vari-
cies in the sites examined. The primary mechanism of native ous parts of the world. Vegetatio, 61, 65–75.
species displacement appears to be burial rather than com- ELSER, A.E., 1970. Manawatu sand dune vegetation. Proceedings of
petition for nutrients. At the sites examined Desmoschoenus the New Zealand Ecological Society, 17, 41–46.
HERTLING, U.M. and LUBKE, R.A., 1999a. Use of Ammophila aren-
displacement is most rapid in situations of active sedimen-
aria for dune stabilization in South Africa and its current distri-
tation and dune development. bution—perceptions and problems. Environmental Management,
There is no evidence that the relatively high tolerance of 24, 467–482.
Desmoschoenus to salt will allow it to occupy the lower stoss HERTLING, U.M. and LUBKE, R.A., 1999b. Indigenous and Ammo-
face of the foredune as observed by PARTRIDGE (1995). Des- phila arenaria-dominated dune vegetation on the South African
Cape coast. Applied Vegetation Science, 2, 157–168.
moschoenus cannot co-exist with Ammophila in the active HESP, P.A., 1981. The formation of shadow dunes. Journal of Sedi-
dune systems of the study areas and is threatened with local mentary Petrology, 51, 101–12.
extermination if Ammophila invasion continues. HESP, P.A., 1988. Morphology, dynamics and internal stratification
of some established foredunes in Southeast Australia. Sedimen-
tary Geology, 55, 17–41.
EPILOGUE HESP, P.A. and THOM, B.G., 1990. Geomorphology and evolution of
The Department of Conservation commenced aerial herbi- active transgressive dunefields. In: NORDSTROM, K.; PSUTY, N.,
and CARTER, B., (eds.), Coastal Dunes: Form and Process. Chich-
cide eradication of Ammophila in Doughboy Bay in January ester: J. Wiley and Sons, 251–283pp.
1999. Ammophila is to be eradicated from all three dune sys- HEYLIGERS, P., 1985. The impact of introduced plants on foredune
tems. This program was extended to include Mason Bay in formation in south-eastern Australia. Proceedings of the Ecological
January 2001. Ammophila has been virtually eradicated from Society of Australia, 14, 23–41.
HOLLAND, L.D., 1981. Plants and sand dune development, Ammo-
the southern dune system in Doughboy Bay after three ap- phila arenaria versus Desmoschoenus spiralis on Kaitorete Bar-
plications of herbicide (1999–2001). A new dune landscape is rier, Canterbury. Unpublished MSc Thesis, Christchurch: Univer-
developing. sity of Canterbury, 141p.
HUISKES, A.H.L., 1979. Biological For a of the British Isles. Am-
mophila arenaria (L.) Link (Psamma arenaria (L.) Roem. et
ACKNOWLEDGMENTS Schult.; Calamagrostis arenaria (L.) Roth). Journal of Ecology, 67,
The authors are grateful for the cooperation and support 363–82.
HULTEN, E. and FRIES, M. 1986. Atlas of North European Vascular
extended by the Department of Conservation, Southland Plants: North of the Tropic of Cancer. I. Introduction. Germany:
Conservancy and Stewart Island Field Center over several Kooltz Scientific Books, 1172p.
years; particularly Chrissy Wickes; Mr Murray Nieuwen- JOHNSON, P.N., 1992. The Sand Dune and Beach Vegetation Inven-
huyse; and Dr Carol West. We would also like to thank Mr tory of New Zealand: II South Island and Stewart Island. Christ-
church: Land Resources Scientific Report No. 16, Department of
Bill Mooney (cartography) and Mr Bill Moffat (Southeast
Scientific and Industrial Research, 278p.
Air). Mike Hilton would like to gratefully acknowledge the MAUN, M.A., 1998. Adaptations of plants to burial in coastal sand
hospitality of the Department of Geography and Environ- dunes. Canadian Journal of Botany, 76, 713–738.
mental Studies at the University of Tasmania and De- PARTRIDGE, T., 1992. The Sand Dune and Beach Vegetation Inventory
partment of Geography and Environmental Studies at the of New Zealand: I North Island. Christchurch: Land Resources Sci-
entific Report No. 16, Department of Scientific and Industrial Re-
University of Adelaide, while on sabbatical leave during search, 253p.
2001. PARTRIDGE, T., 1995. Interaction Between Pingao and Marram on
Sand Dunes. Wellington: Department of Conservation, 27p.
LITERATURE CITED PICKART, A.J. and SAWYER, J.O., 1998. Ecology and restoration of
Northern California coastal dunes. California: California Native
BAYE, P.R., 1990. Comparative Growth Responses and Population Plant Society, 152p.
Ecology of European and American Beachgrasses (Ammophila spp.) PYE, K., 1983. Coastal dunes. Progress in Physical Geography, 7,
in Relation to Sand Accretion and Salinity. Doctoral Dissertation, 531–557.
University of Western Ontario, London, Ontario, 324p. RANWELL, D.S., 1972. Ecology of Salt Marshes and Sand Dunes.
BUELL, A.C.; PICKART, A.J., and STUART, J.D., 1995. Introduction London: Chapman and Hall, 258p.
history and invasion patterns of Ammophila arenaria on the North REID, V.A., 1998. The impact of weeds on threatened plants. Science
Coast of California. Conservation Biology, 9, 1587–1593. and Research Internal Report Number 164. Wellington: Depart-
CASTRO, C.A., 1988. The artificial construction of foredunes and the ment of Conservation, 67p.
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 185
SYKES, M.T. and WILSON, B., 1990a. An experimental investiga- WIEDEMANN, A.M. and PICKART, A., 1996. The Ammophila problem
tion into the response of New Zealand sand dune species to dif- on the Northwest coast of North America. Landscape and Urban
ferent depths of burial by sand. Acta Botanica Neelandica, 39, Planning, 34, 287–299.
171–181. WILLIS, A.J.; FOLKES, B.F.; HOPE-SIMPSON, J.F., and YEMM, E.W.,
SYKES, M.T. and WILSON, B., 1990b. Dark tolerance in plants of 1959. Braunton Barrows: the dune system and its vegetation. II.
dunes. Functional Ecology, 4, 799–805. Journal of Ecology, 47, 249–288.
Journal of Coastal Research, Vol. 21, No. 1, 2005
Processes of Ammophila arenaria (Marram Grass)
Invasion and Indigenous Species Displacement,
Stewart Island, New Zealand
Mike Hilton†, Megan Duncan‡, and Anne Jul§
†Department of Geography ‡Department of Conservation §Otago Regional Council
University of Otago Private Bag 68-908 70 Stafford St.
PO Box 56 Newton, Auckland, Dunedin, New Zealand
Dunedin, New Zealand New Zealand
ABSTRACT
HILTON, M.; DUNCAN, M., and JUL, A., 2005. Processes of Ammophila arenaria (marram grass) invasion and
indigenous species displacement, Stewart Island, New Zealand. Journal of Coastal Research, 21(1), 175–185. West
Palm Beach (Florida), ISSN 0749-0208.
The present study (1) describes the rates and patterns of Ammophila arenaria (marram grass) invasion in a large
transgressive dune system (Mason Bay) and on a prograding foredune-ridge barrier (Doughboy Bay), Stewart Island,
New Zealand; (2) examines the impact of Ammophila on dune morphology and indigenous dune biota; and (3) assesses
the significance of geomorphic processes in accounting for the patterns observed. Processes of Ammophila invasion
are interpreted from evidence of landform development and vegetation change; field observations and survey of dune
landforms and dune vegetation; the aerial photographic record and historic accounts of the local botany.
The area dominated by Ammophila in the Mason Bay study area has increased from 1.4 ha in 1958, to 17.8 ha in
1978, to 74.9 ha in 1998; a 5,204 percent increase. Ammophila invasion of active dune systems in the study areas is
clearly associated with dune forming processes—shadow dune development; migration of long-walled parabolic dunes;
stoss face blowout development; and barrier progradation. The primary mechanism of native species displacement
appears to be burial rather than competition for nutrients. Ammophila traps sand and builds dunes at rates that may
exceed the threshold of tolerance of local native species. Desmoschoenus spiralis, the dominant indigenous foredune
species, cannot co-exist with Ammophila in the active dune systems investigated.
ADDITIONAL INDEX WORDS: Desmoschoenus spiralis, foredune complex, parabolic dune migration, Mason Bay,
Doughboy Bay, burial, shadow dune.
INTRODUCTION To date, the processes of Ammophila invasion and indige-
nous plant displacement have only been described in general
Ammophila arenaria (Marram Grass or European Beach-
terms. For example, BUELL, PICKART and STUART (1995,
grass) has been introduced to the active dune systems of
p.1591) attribute the rapid and successful spread of Ammo-
southeast Australia (HESP and THOM, 1990; CULLEN, 1998);
phila through the dunes of North Spit, Humboldt Bay, Cali-
North America (WIEDEMANN and PICKART, 1996); South Af-
fornia, to ‘‘a number of processes and conditions: optimal hab-
rica (HERTLING and LUBKE, 1999a); Chile (HULTEN and
FRIES, 1986; CASTRO, 1988) and New Zealand (JOHNSON, itat, multiple introduction, natural and human disturbance,
1992; PARTRIDGE, 1992). At most sites it has been planted to and proximity to the strand where rhizome fragments are
construct or re-establish foredunes or stabilise transgressive washed ashore by storm surf’’. Increasing our understanding
dune systems. of these processes will assist conservation agencies assess the
Ammophila has belatedly been shown to be highly invasive vulnerability of dune systems to Ammophila invasion and in-
and a threat to the ecology of active dune systems outside its crease the effectiveness of Ammophila eradication operations.
natural range. It may adversely affect indigenous dune flora The impact of Ammophila on the native dune flora and
and the diversity of habitats by stabilizing naturally mobile fauna of New Zealand has been recognised for some time (e.g.
dunes and accelerating vegetation succession (COOPER, 1958; JOHNSON, 1992; DEPARTMENT OF CONSERVATION, 1997).
BUELL, PICKART and STUART, 1995; HERTLING and LUBKE, New Zealand has three indigenous foredune sandbinders,
1999b). The area of bare sand, species diversity and evenness Desmoschoenus spiralis (a sedge), Austrofestuca littoralis (a
(uniformity in abundance and cover of species) has been perennial grass) and Spinifex sericeus (a perennial grass).
shown to correlate negatively with Ammophila along the west Desmoschoenus and Austrofestuca occur throughout New Zea-
coast of the United States (reviewed by PICKART and SAW- land while Spinifex occurs throughout the North Island and
YER, 1998) and in Tasmania (reviewed by CULLEN, 1998). in northern South Island. Ammophila has displaced Desmos-
choenus in many dune systems, but particularly around the
01041 received 3 May 2001; accepted in revision 25 July 2002. south and east coasts of the South Island and the exposed
176 Hilton, Duncan and Jul
western coasts of the three main islands. Ammophila is al- Bay dunefield is dominated by parabolic dune forms. Am-
most the only foredune species on exposed western coasts and mophila was planted near Kilbride at the southern end of
around the south coast of the South Island of New Zealand. Mason Bay in the early 1930s (Figure 1) and east of the
A recent inventory of threatened plant species in New Zea- ‘Homestead’ in the mid 1960s (Figure 2). Ammophila now
land concluded Ammophila was a threat to ten indigenous dominates the foredune environment but is patchy and pre-
dune species, including Desmoschoenus and Austrofestuca sent in very low densities across approximately 60 per cent
(REID, 1998). of the hinterland of the active dunefield between Martin’s
The displacement of indigenous dune plant species in mo- Creek and Duck Creek. Austrofestuca littoralis and Desmos-
bile dune systems appears to be related to geomorphic pro- choenus remain the primary sandbinders landward of the
cesses of dune development. Ammophila-dominated dunes stonefield.
are generally steeper and higher than dunes formed in as-
sociation with the indigenous foredune species, Desmoschoen-
Doughboy Bay
us spiralis and Spinifex sericeus in New Zealand (ESLER,
1970; HOLLAND, 1981). PARTRIDGE (1995) concluded Am- The active dune systems of Doughboy Bay provide oppor-
mophila can either co-exist with Desmoschoenus or displace tunities to study processes of marram grass invasion on a
it, depending on several factors. The species can co-exist in rapidly prograding bay-head barrier. Until recently the series
three situations: (1) on the front of the foredune where mar- of foredune ridges that comprise the seaward half of the
ram grass spread is limited by high salt concentrations; (2) southern barrier had not experienced significant disturbance.
where moisture in the upper sand layers is not limiting; and These are primary dune forms that have developed in con-
(3) where both species are moribund because of low rates of junction with Ammophila. The foredune ridges formed since
sand accumulation. Displacement of Desmoschoenus was the 1960s, probably as a result of post-storm sedimentation.
found to be most severe on stable dunes and where moisture
in the upper layers of sand was limiting. Desmoschoenus dis-
placement was found to be unrelated to the form and geo- METHODS
morphic processes of different kinds of dunes.
The process of Ammophila dispersal and invasion in active
Patterns of Ammophila Invasion
dune systems is not well documented in New Zealand, partly
The distribution and density of Ammophila in the Mason
because most dune systems were completely occupied by Am-
Bay study area were mapped from vertical aerial photo-
mophila by the mid 1900s. Specific geomorphic conditions
graphs flown in 1958, 1978 and 1998. The 1958 photographs
that aid or hinder Ammophila invasion have not been de-
are relatively poor black and white images, but of sufficient
scribed. The present study (1) describes the rates and pat-
quality to allow the identification of areas of Ammophila. The
terns of Ammophila invasion at two sites on the west coast
remainder are high quality colour images, printed at 1:3,000
of Stewart Island, New Zealand; (2) evaluates the importance
and 1:10,000. Resolution of contrasting features is approxi-
of geomorphic processes in accounting for these patterns; and
mately 1 m2. The southern dune system of Doughboy Bay was
(3) interprets the rate and processes of Desmoschoenus dis-
photographed in 1977, 1987 and 1999. The 1977 Doughboy
placement by Ammophila.
image is also black and white. Maps of Ammophila distribu-
STUDY AREAS tion and density derived from the 1998 (Mason Bay) and 1999
(Doughboy Bay) aerials were field checked during 1999.
Mason Bay Maps of the location and density of Ammophila between
The combination of available sediment and prevailing on- Duck Creek and Martin’s Creek were constructed using ARC/
shore winds has led to extensive dune building and trans- INFO software. The 1:3,000 1958 image was digitised by
gressive dunefield development along the western and north- hand. The 1978 and 1998 images were scanned at 1:10,000.
ern coasts of Stewart Island. Prevailing winds are from the This permitted on-screen digitising of these images using the
west (Figure 1). The active dune systems of Stewart Island 1:3,000 images for reference. All Mason Bay maps were or-
encompass a range of primary and secondary dune forms in tho-rectified based on GPS observations of distinctive fea-
a variety of depositional environments. Mason Bay contains tures.
the largest transgressive dune system on Stewart Island and Five density classes of Ammophila were mapped in the Ma-
one of the largest and least modified dune systems in New son Bay study area (0–5, 6–25, 26–50, 51–75 and greater
Zealand. Between Martin’s Creek and Duck Creek, the active than 75 percent). Classes refer to the proportion of ground
dune system extends up to 3 km inland. The foredune com- covered by Ammophila within discrete areas. These areas
plex is a large, continuous feature up to 15 m high and 150 generally correspond with the boundaries of particular land-
m wide. Landward of the foredune a series of active U- forms, for example, the depositional lobes of parabolic dunes.
shaped, long-walled parabolic dunes (after PYE, 1983; HESP Ammophila is considered the ‘dominant’ species where it com-
and THOM, 1990) are transgressing a broad, gently-sloping prises more than 50 percent ground cover within a discrete
stonefield. These dunes are evident in 1958 aerial photo- area. Just one density class is recognised at Doughboy Bay
graphs and formed prior to the arrival of Ammophila in the ( 50 per cent). Ammophila occupied 100 percent of the active
central area of Mason Bay. The hinterland of the dune system dune system seaward of the storm scarp in 1977 at densities
north of Martin’s Creek to the northern end of the Mason exceeding 50 percent ground cover.
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 177
Figure 1. Location of study sites and major Stewart Island active sand dune systems. Wind rose derived from hourly observations, Southeast Cape,
Stewart Island (1992–97).
Landform Development and Vegetation Change dune complex (40); (iii) the deflation zone of the long-walled
parabolic dunes (40); (iv) trailing arms (40); and (v) deposi-
Contemporary vegetation cover in the Mason Bay study tional lobe (25). The average ground cover for each stratum
area was estimated by random sampling (1 m2 quadrat) in was derived by averaging the percent cover of the major spe-
January 2000. The quadrat data was then stratified into five cies observed in each quadrat. Our interpretation of species
classes based on geomorphic characteristics: (i) the stoss face diversity at Doughboy Bay is based on data gathered over a
of foredune complex (40 sites); (ii) the lee slope of the fore- three year period (1999–2001) from ten permanent quadrats
Journal of Coastal Research, Vol. 21, No. 1, 2005
178 Hilton, Duncan and Jul
Mason Bay in 1908, prior to the introduction of Ammophila
(COCKAYNE, 1909). In addition, Ammophila-dominated fore-
dunes in the Mason Bay study area were compared with rem-
nant foredunes in northern Mason Bay. Desmoschoenus is
still the primary sandbinder along small stretches of coast-
line north of Duck Creek and is the dominant sandbinder
across large areas of the hinterland.
RESULTS
Patterns of Ammophila Invasion
Ammophila spread north in Mason Bay following introduc-
tion at Kilbride by farmers in the 1930s. By 1958 Ammophila
had established a small number of dense colonies north of
Martin’s Creek (Figure 2). The total area of dense Ammophi-
la, where Ammophila cover exceeded 50 percent, was around
1.4 ha. At that time almost all Ammophila occurred within
400 m of the toe of the foredune. Further north, beyond the
advancing Ammophila, the foredune comprised a band of low
shadow and/or coppice dunes, approximately 40 m wide. The
indigenous foredune vegetation in 1958 was sparse (compris-
ing, perhaps, 10–30 per cent surface cover). Desmoschoenus,
Austrofestuca and Euphorbia glauca were probably common
at foredune sites and backdune sites experiencing high rates
of sedimentation. Euphorbia glauca was described by COCK-
AYNE (1909) as common and was still relatively abundant in
the 1960s. Pimelea lyallii (Sand Daphne), Coprosma acerosa
(Sand Coprosma), Gentiana saxosa and Raoulia hookeri were
probably also widespread in relatively sheltered backdune
and deflation areas. The long-walled parabolic dunes de-
scribed above lay to landward of this foredune environment
in 1958.
The extent and density of Ammophila increased between
1958 and 1978 (Figure 2). By 1978, Ammophila was present
in dense patches up to 750 m inland. A continuous band of
Ammophila occupied the former foredune environment. The
width and density of this band decreased towards Duck
Creek. A continuous, though topographically irregular fore-
dune was present (Stage III after HESP, 1988). The extent
and density of Ammophila in the study area continued to in-
crease between 1978 and 1998. During this period Ammo-
phila also spread from the 1960s plantings (east of the Home-
stead) about 700 m along the north-eastern margin of the
Figure 2. Ammophila arenaria invasion of Mason Bay study area, 1958–
1998, based on aerial photography interpretation. Only the 1998 pattern dune system (Figure 2).
was field checked. The proportion of the study area dominated by Ammophila
increased from 1.4 ha in 1958 to 17.8 ha in 1978 to 74.9 ha
in 1998. Therefore, the area dominated by Ammophila in-
creased 5,204 per cent in the period 1958–1998. The area
located across the Ammophila-dominated section of the with little or no Ammophila (0–5 percent ground cover) de-
southern dune barrier (which comprises about half of the to- clined 46 per cent over the same period.
tal area of the barrier). These quadrats were established to Available aerial photographs show that Ammophila spread
monitor vegetation change during an Ammophila eradication from the south in Mason Bay (taking approximately 25 years
programme, which commenced in January 1999. to form a continuous foredune between Martins Creek and
Historic vegetation cover (1908–present) and associated Duck Creek). Ammophila invasion of the transgressive dune
landforms were interpreted from aerial photographs, a sur- system in Masons Bay is continuing. In contrast, Ammophila
vey of dune landforms and dune biota (including dead plant became the dominant foredune and backdune species at the
material, especially Desmoschoenus root material); and early Doughboy Bay site over a much shorter period. The area of
botanical accounts. Leonard Cockayne, one of New Zealand’s Ammophila increased rapidly between 1977 and 1999 as the
pioneering botanists, described the dunes and dune flora of barrier prograded by the formation of foredune ridges, from
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 179
Landform Development and Vegetation Change
Since the arrival of Ammophila the foredune in central Ma-
son Bay study area has undergone considerable change.
COCKAYNE (1909) described aspects of the pre-Ammophila
dune landscape and dune flora of Mason Bay. In 1908 the
foredune comprised a line of ‘‘low hummocks’’, 2–3 m high.
The probable position and elevation of these dunes relative
to the modern foredune complex is shown in Figure 4. COCK-
AYNE’s description of the foredune is consistent with the dune
landscape as can be discerned from the 1958 aerial photo-
graphs, soon after Ammophila established north of Martins
Creek. The aerial photographs indicate a hummocky, discon-
tinuous foredune, probably equivalent to the Stage IV fore-
dune of HESP (1988).
The foredune complex developed between 1958 and 1978
as adjacent shadow dunes coalesced and Ammophila estab-
lished a semi-uniform vegetation cover. The foredune is now
a relatively massive landform termed a ‘foredune complex’
(after DOING, 1985). Deposition of sand across the Mason Bay
foredune complex has continued as a result of: (1) formation
of minor blowouts affecting the stoss face of the foredune; and
(2) erosion and transport of sands across the beach during
episodes of strong onshore winds. During the latter, sand is
transported and deposited across the width of the foredune
complex, well inland of the crest. Both processes appear to
contribute to the vertical accretion and lateral growth of the
foredune complex. Between 1978 and 1998 the morphology of
the stoss face of the foredune complex has become more reg-
ular as the density of Ammophila increased and the extent
and frequency of foredune blowouts declined. Narrow blow-
outs of a few metres wide and 20 m or so deep still occur, but
these appear short-lived and are minor compared with the
mass of the foredune complex. Periodic scarping during storm
conditions limits the growth of Ammophila down the lower
slopes of the stoss face of the foredune complex; hence, the
plan configuration of the seaward edge of the foredune com-
plex has become very regular since the establishment of Am-
mophila (Figure 2).
The long-walled parabolic dunes in the central Mason Bay
study area developed before Ammophila invasion—they are
present in the 1958 aerial photographs and were briefly de-
scribed by COCKAYNE (1909). They are active transgressive
dunes. Their depositional lobes have migrated at average
rates of 5.0–7.5 m per year since 1978. The morphology and
Figure 3. Morphology and plant cover of parabolic dune number six (in plant cover of parabolic dune number six is typical of these
1998). The contemporary dimensions of the foredune complex are com-
pared with Cockayne’s (1909) description of the foredune in 1908.
landforms. Four morphologic components are recognisable:
(1) the deflation zone; (2) an area of active erosion incorpo-
rating the ‘throat’; (3) a depositional lobe comprising a com-
plex of coppice and shadow dunes; and (4) shared trailing
about 1.7 ha in 1977 to about 7.0 ha in 1999, a 412 per cent arms (Figure 4). These elements of the parabolic dune have
increase. At the same time the barrier prograded seawards been maintained as the throat and depositional lobe ad-
(45–60 m) and towards the north (140 m) (Figure 3). The vanced inland and the trailing arms lengthened in the period
foredune ridges formed since 1977 have been colonised by 1978 to 1998.
Ammophila and now Ammophila is the only species inhabit- Live Desmoschoenus was observed during 1999 in the de-
ing the stoss face of the modern foredune. There is no evi- flation zones of most of the parabolic dunes in the study area,
dence, exposed rhizome for example, that Desmoschoenus has (although most was moribund); on the level surfaces of the
ever formed a significant cover on the foredune ridges, trailing arms (with Ammophila and a range of native dune
though a handful of isolated plants were observed in 1999. and opportunistic species); and, occasionally, forming isolated
Journal of Coastal Research, Vol. 21, No. 1, 2005
180 Hilton, Duncan and Jul
Figure 4. Plant species associated with the foredune complex (stoss and backdune slopes), trailing arms, deflation zones and depositional lobes. Values
based on random survey of all long-walled parabolic dunes between Duck Creek and Martins Creek.
shadow dunes in the depositional lobes. Aerial photography bly present (Figure 6b), or Ammophila arrived soon after the
(1958) and the widespread occurrence of dead root material storm. Subsequent, post-storm, progradation occurred in
in the eroding faces of the trailing arms and across the throat conjunction with Ammophila, not Desmoschoenus (Figure 6c
area shows that Desmoschoenus was, until recently, the & 6d) as successive foredune ridges established seawards of
dominant sand-binding species on these dunes. These dunes the former storm scarp. Ammophila now forms a dense cover
are now evolving in conjunction with Ammophila, which ac- across the stoss face of the contemporary foredune and a
counts for about half the plant cover across the trailing relatively sparse cover further inland. Desmoschoenus is ab-
arms, less than half the total plant cover in the deflation sent from the foredune ridge section of the barrier. The
zones and over 90 percent of the plant cover in the deposi- storm scarp, therefore, separates the older Desmoschoenus
tional lobes (Figure 5). barrier from the foredune ridge landscape associated with
The pre-Ammophila barrier in Doughboy Bay probably Ammophila. Ammophila and Desmoschoenus occur land-
comprised a relatively simple foredune complex, 100–150 m ward of the scarp, where Gaultheria macrostigma and other
wide (Figure 6a). Some time in the mid 1900s (ca. 1960) an creeping species form a dense mat and where sand move-
exceptional storm event eroded the front of the barrier. At ment is nil. Neither species will survive if stability contin-
this time both Desmoschoenus and Ammophila were proba- ues.
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 181
Figure 5. Recent progradational history of the southern dune system, Doughboy Bay, 1977–1999. Progradation has occurred in conjunction with Am-
mophila arenaria. Desmoschoenus spiralis is now restricted to the remnant barrier landward of the storm scarp.
Figure 6. Interpretation of coastal change and Ammophila invasion, southern dune system, Doughboy Bay, ca. 1900–1999.
Journal of Coastal Research, Vol. 21, No. 1, 2005
182 Hilton, Duncan and Jul
DISCUSSION Creek and Duck Creek. Over the same period, Desmoschoenus
and Austrofestuca were displaced from the foredune and gen-
Ammophila Dispersal and Invasion erally lost from the adjoining parabolic dunes. These pro-
The establishment of Ammophila in the Mason Bay study cesses of foredune development and indigenous foredune spe-
area probably resulted from clonal spread, since reproduction cies displacement are ongoing north of Duck Creek. Austro-
by seed is thought to be exceptional (HUISKES, 1979). Frag- festuca littoralis and Euphorbia glauca, described by COCK-
ments of rhizome are tolerant of seawater rafting for at least AYNE as common in 1908, are now also absent from the
8 tidal cycles (BAYE, 1990) and probably much longer (e.g. foredune and parabolic dune environment in the Mason Bay
HEYLIGERS, 1985). Fragments of rhizome could have entered study area.
the sea in southern Mason Bay, during foredune scarping; Burial is probably the principle mechanism of indigenous
and then been transported alongshore by nearshore currents species displacement. Individual Ammophila plants respond
and into the foredune environment by storm wave swash. to burial by rapid production of elongated stem internodes
This process may have occurred during a single storm event (HUISKES, 1979). Shoots develop from these nodes, which
or phase of storminess, since most of the patches of Ammo- then produce leaves upon reaching the surface. This process
phila in 1958 in the study area appear of similar age (3–5 is repeated if sand supply continues, eventually producing a
years). tussock habit (CHAPMAN, 1964). These tussocks create eddies
Ammophila is most likely to have established in the inter- or vortices in the lee of the plant during phases of onshore
dune hollows of the study area, some distance inland of the sand transport that tend to produce a pyramidal-shaped
mean high tide line, following deposition of rhizomes during shadow dune (HESP, 1981). Ammophila shadow dune devel-
storm conditions. Storm-wave deposition of rhizome propa- opment results in areas of rapid accretion. As individual
gules would have been assisted by the discontinuous foredune dunes develop intervening depressions may experience in-
as it existed in 1958. The contemporary Desmoschoenus fore- tense erosion. Sand deposited in the lee of these dunes, or as
dunes north of Duck Creek are typically low shadow dunes depositional lobes downwind of interdune deflation areas,
2–4 m high, with multiple interdune openings at about the may extend the influence of the first line of Ammophila shad-
level of the spring high tide line. Interdune hollows in the ow dunes well inland (Figure 7a). Desmoschoenus located in
foredune are flooded during storm events and commonly the lee of the initial Ammophila plant and in the new inter-
choked with flotsam and wind-blown material. dune hollows may experience lethal rates of accretion and
Successful Ammophila dispersal may rely on extreme erosion, respectively.
storm events that are able to both erode propagules from the These processes may, over time, establish a massive fore-
face of the foredune and deposit them in suitable locations. dune complex where none existed previously. Subsequent
Ammophila is intolerant of substrate salt concentrations in blowout development across the stoss face may contribute to
excess of 1–1.5 percent (HUISKES, 1979; CHAPMAN, 1964) and the vertical accretion of the foredune complex (Figure 7b),
may have difficulty establishing low on the stoss face of Des- which may also work to maintain Ammophila vigor. All plant
moschoenus foredunes, in circumstances where inundation is species experience stress in the deflation areas of these blow-
common. Ammophila rhizomes blown or washed well inland, outs, however, Ammophila plants on the margins of the de-
beyond the usual reach of storm waves, may stand the best flation basin and across the depositional lobes experience re-
chance of survival. Successful establishment of Ammophila newed growth following accelerated sand deposition. Subse-
may also depend on favourable meteorological conditions fol- quent accretion and recolonisation by Ammophila may rap-
lowing transport and deposition of rhizomes. Ammophila es- idly close the blowout.
tablishment, therefore, may be episodic, even where the sup- Ammophila invasion and concomitant dune development in
ply of rhizome material is constant. the Mason Bay study area appears to result in rates of sand
Two other factors may have affected the potential for Am- accretion (and erosion) that exceed the tolerance of native
mophila to establish on or close to the stoss face of the former dune species. Ammophila is known to accelerate rates of sand
foredune in Mason Bay. Compared with coasts in southeast accretion (CHAPMAN, 1964) and burial is recognised as a
Australia, Europe and elsewhere, the study site lacks her- strong selective force in dune ecology (MAUN, 1998). The abil-
baceous pioneer annuals (the ‘ephemerous tidemark’ and ‘pe- ity of a plant to survive burial includes the plants tolerance
rennial tidemark’ communities of DOING (1985)). Species of darkness and capacity to grow up through sand. Ammo-
such as Cakile maritime may provide suitable conditions for phila is able to tolerate rates of sand accumulation of up to
Ammophila establishment across incipient foredunes but 1 m per year (WILLIS et al., 1959; RANWELL, 1972), although
have only recently established in the study area. Secondly, experimental work indicates instantaneous total burial will
berm development is unusual along Mason Bay beach and kill most specimens (SYKES and WILSON, 1990a). However,
waves frequently cut the toe of the foredune. SYKES and WILSON (1990b) have demonstrated that Ammo-
phila is much more tolerant of darkness (surviving burial for
Indigenous Species Displacement 117 days) than the native foredune species; Austrofestuca (37
days); Euphorbia (50 days); or Desmoschoenus (67 days). Am-
Displacement of Desmoschoenus and Austrofestuca occurred mophila may possess other competitive advantages (tolerance
rapidly following Ammophila invasion in central Mason Bay of drought, for example), but these are probably of minor im-
and Doughboy Bay. Between 1958 and 1978 Ammophila es- portance compared to the ability of this species to initiate and
tablished a single, continuous foredune between Martin’s survive rapid sand accumulation.
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 183
Figure 7. Interpretations of landform development and Ammophila invasion in the study areas.
Displacement of Desmoschoenus across the long-walled the survival threshold of these species. At the same time
parabolic dunes in Mason Bay has also been rapid since the the indigenous species in the depositional lobe and throat
establishment of Ammophila. This process is an inevitable zones are destroyed as the dune advances downwind. The
consequence of dune migration. The throat and depositional trailing arms are relatively stable and hence retain higher
lobes are eroded as the dunes advance. The eroded sand is indigenous species diversity. Secondly, where Desmoschoen-
transported downwind and landward, usually through a us is able to establish in the depositional lobes, it is probably
narrow throat, or multiple throats, to contribute to the pro- displaced by burial.
gressive construction of a new depositional lobe, progres- Finally, Ammophila invasion may be very rapid on
sively landward of the former (Figure 7c). In this fashion, coasts experiencing progradation (Figure 7d). Associated
the depositional lobes are being continually recycled. The loss of native dune species may be equally rapid. Ammo-
destruction of Desmoschoenus is, therefore, a natural pro- phila colonized developing foredunes at Doughboy Bay fol-
cess. The failure of Desmoschoenus to re-establish in the de- lowing a significant storm event more rapidly than Des-
positional lobes and trailing arms is a consequence of com- moschoenus. In this case Ammophila invasion and Desmos-
petition with Ammophila. Ammophila appears better able choenus displacement may be very rapid, of the order of
and more aggressive at colonising the depositional lobes of years. In such dynamic circumstances inter-specific com-
these landforms. Once established, Ammophila may dis- petition for nutrients is unlikely to be a significant dis-
place indigenous species by accelerating accretion beyond placement mechanism.
Journal of Coastal Research, Vol. 21, No. 1, 2005
184 Hilton, Duncan and Jul
CONCLUSIONS interference of dune-beach interaction, Chile. Journal of Coastal
Research Special Issue, 3, 103–107.
In summary, Ammophila invasion and Desmoschoenus dis- CHAPMAN, V.J., 1964. Coastal Vegetation. London: Pergamon Press,
placement in Mason Bay and at Doughboy Bay is associated 245p.
COCKAYNE, L., 1909. Report on a Botanical Survey of Stewart Island.
with at least four processes of landform development (Figure Wellington: Government Printers, 66p.
7): (a) shadow dune and foredune development; (b) blowout COOPER, W.S., 1958. Coastal Sand Dunes of Oregon and Washing-
development; (c) parabolic dune migration; and (d) barrier ton. Memoir 72, Geological Society of America, Boulder, Colora-
progradation. do, 85p.
CULLEN, P., 1998. Ammophila arenaria and Euphorbia paralias: Se-
Ammophila invasion of the active dune systems examined
rious Threats to the Integrity of the South West Tasmanian Coast-
is clearly associated with dune forming processes. Competi- line. Unpublished Report, Hobart: Tasmanian Parks and Wildlife
tion for nutrients is almost certainly important in semi-stable Service, 18p.
situations, but not in dynamic geomorphic situations. In such DEPARTMENT OF CONSERVATION, 1997. Ecology and Management of
circumstances Ammophila traps sand and builds dune at Invasive Weeds. Conservation Sciences Publication No. 7. Welling-
ton: Department of Conservation, 67p.
rates that exceed the threshold of tolerance of all native spe- DOING, H., 1985. Coastal foredune zonation and succession in vari-
cies in the sites examined. The primary mechanism of native ous parts of the world. Vegetatio, 61, 65–75.
species displacement appears to be burial rather than com- ELSER, A.E., 1970. Manawatu sand dune vegetation. Proceedings of
petition for nutrients. At the sites examined Desmoschoenus the New Zealand Ecological Society, 17, 41–46.
HERTLING, U.M. and LUBKE, R.A., 1999a. Use of Ammophila aren-
displacement is most rapid in situations of active sedimen-
aria for dune stabilization in South Africa and its current distri-
tation and dune development. bution—perceptions and problems. Environmental Management,
There is no evidence that the relatively high tolerance of 24, 467–482.
Desmoschoenus to salt will allow it to occupy the lower stoss HERTLING, U.M. and LUBKE, R.A., 1999b. Indigenous and Ammo-
face of the foredune as observed by PARTRIDGE (1995). Des- phila arenaria-dominated dune vegetation on the South African
Cape coast. Applied Vegetation Science, 2, 157–168.
moschoenus cannot co-exist with Ammophila in the active HESP, P.A., 1981. The formation of shadow dunes. Journal of Sedi-
dune systems of the study areas and is threatened with local mentary Petrology, 51, 101–12.
extermination if Ammophila invasion continues. HESP, P.A., 1988. Morphology, dynamics and internal stratification
of some established foredunes in Southeast Australia. Sedimen-
tary Geology, 55, 17–41.
EPILOGUE HESP, P.A. and THOM, B.G., 1990. Geomorphology and evolution of
The Department of Conservation commenced aerial herbi- active transgressive dunefields. In: NORDSTROM, K.; PSUTY, N.,
and CARTER, B., (eds.), Coastal Dunes: Form and Process. Chich-
cide eradication of Ammophila in Doughboy Bay in January ester: J. Wiley and Sons, 251–283pp.
1999. Ammophila is to be eradicated from all three dune sys- HEYLIGERS, P., 1985. The impact of introduced plants on foredune
tems. This program was extended to include Mason Bay in formation in south-eastern Australia. Proceedings of the Ecological
January 2001. Ammophila has been virtually eradicated from Society of Australia, 14, 23–41.
HOLLAND, L.D., 1981. Plants and sand dune development, Ammo-
the southern dune system in Doughboy Bay after three ap- phila arenaria versus Desmoschoenus spiralis on Kaitorete Bar-
plications of herbicide (1999–2001). A new dune landscape is rier, Canterbury. Unpublished MSc Thesis, Christchurch: Univer-
developing. sity of Canterbury, 141p.
HUISKES, A.H.L., 1979. Biological For a of the British Isles. Am-
mophila arenaria (L.) Link (Psamma arenaria (L.) Roem. et
ACKNOWLEDGMENTS Schult.; Calamagrostis arenaria (L.) Roth). Journal of Ecology, 67,
The authors are grateful for the cooperation and support 363–82.
HULTEN, E. and FRIES, M. 1986. Atlas of North European Vascular
extended by the Department of Conservation, Southland Plants: North of the Tropic of Cancer. I. Introduction. Germany:
Conservancy and Stewart Island Field Center over several Kooltz Scientific Books, 1172p.
years; particularly Chrissy Wickes; Mr Murray Nieuwen- JOHNSON, P.N., 1992. The Sand Dune and Beach Vegetation Inven-
huyse; and Dr Carol West. We would also like to thank Mr tory of New Zealand: II South Island and Stewart Island. Christ-
church: Land Resources Scientific Report No. 16, Department of
Bill Mooney (cartography) and Mr Bill Moffat (Southeast
Scientific and Industrial Research, 278p.
Air). Mike Hilton would like to gratefully acknowledge the MAUN, M.A., 1998. Adaptations of plants to burial in coastal sand
hospitality of the Department of Geography and Environ- dunes. Canadian Journal of Botany, 76, 713–738.
mental Studies at the University of Tasmania and De- PARTRIDGE, T., 1992. The Sand Dune and Beach Vegetation Inventory
partment of Geography and Environmental Studies at the of New Zealand: I North Island. Christchurch: Land Resources Sci-
entific Report No. 16, Department of Scientific and Industrial Re-
University of Adelaide, while on sabbatical leave during search, 253p.
2001. PARTRIDGE, T., 1995. Interaction Between Pingao and Marram on
Sand Dunes. Wellington: Department of Conservation, 27p.
LITERATURE CITED PICKART, A.J. and SAWYER, J.O., 1998. Ecology and restoration of
Northern California coastal dunes. California: California Native
BAYE, P.R., 1990. Comparative Growth Responses and Population Plant Society, 152p.
Ecology of European and American Beachgrasses (Ammophila spp.) PYE, K., 1983. Coastal dunes. Progress in Physical Geography, 7,
in Relation to Sand Accretion and Salinity. Doctoral Dissertation, 531–557.
University of Western Ontario, London, Ontario, 324p. RANWELL, D.S., 1972. Ecology of Salt Marshes and Sand Dunes.
BUELL, A.C.; PICKART, A.J., and STUART, J.D., 1995. Introduction London: Chapman and Hall, 258p.
history and invasion patterns of Ammophila arenaria on the North REID, V.A., 1998. The impact of weeds on threatened plants. Science
Coast of California. Conservation Biology, 9, 1587–1593. and Research Internal Report Number 164. Wellington: Depart-
CASTRO, C.A., 1988. The artificial construction of foredunes and the ment of Conservation, 67p.
Journal of Coastal Research, Vol. 21, No. 1, 2005
Ammophila arenaria Invasion in New Zealand 185
SYKES, M.T. and WILSON, B., 1990a. An experimental investiga- WIEDEMANN, A.M. and PICKART, A., 1996. The Ammophila problem
tion into the response of New Zealand sand dune species to dif- on the Northwest coast of North America. Landscape and Urban
ferent depths of burial by sand. Acta Botanica Neelandica, 39, Planning, 34, 287–299.
171–181. WILLIS, A.J.; FOLKES, B.F.; HOPE-SIMPSON, J.F., and YEMM, E.W.,
SYKES, M.T. and WILSON, B., 1990b. Dark tolerance in plants of 1959. Braunton Barrows: the dune system and its vegetation. II.
dunes. Functional Ecology, 4, 799–805. Journal of Ecology, 47, 249–288.
Journal of Coastal Research, Vol. 21, No. 1, 2005