Around the world about equal proportions of sandy coasts can be categorised as stationary, prograding or retrograding (Luijendijk et al. 2018). Within these categories, coastal dunes vary in their behaviours. Dunes may be fixed, in which case their vegetation is generally of later seral species at dense covers, with little to no bare sand visible (e.g., Kenfig dunes in Wales; Pye and Blott 2017), or open, with higher sand cover and early seral species (e.g., Cole Creek, New Zealand; Sykes and Wilson 1991), or actually mobile, as especially occurs on prograding coasts (e.g., Manawatū, New Zealand; Esler 1969; Abhar et al. 2015; Villacís Lozada and Rapson 2021).

Dune mobility can be aggressive, e.g., Holland (1983) recorded rates of inland movement of dunes exceeding 1 km per year. It can also be transgressive, where sand leaves the dunefield to impact on other land covers and/or activities (Esler 1970; Hesp 2001). This may result in managerial intervention to deliberately stabilise sand (e.g., Avis 1989; McKelvey 1999), often with little regard to the conservation values of native vegetation (Williams and West 2000; Gross et al. 2013; Gao et al. 2020; Brandt et al. 2021). Stabilisation has been such that Gao et al. (2020) reported 93% of mobile dunelands reviewed globally had experienced a loss of bare sand over the last 1.5 centuries. In New Zealand 70% of active dunes has been lost, reducing their area to about 40,000 ha (Hilton 2006).

Dunes can be stabilised by a variety of geo-technical methods, including emplacing structures, e.g., groynes, seawalls, break-walls and sand-fences (Nordstrom and Jackson 2013), or by geo-engineering, such as shaping foredunes and beach replenishment (Nordstrom 2014), or even by chemically forming stabilising soil matrices (e.g., Ghadr et al. 2022). A further technique, bioengineering, involves the planting of vegetation, often alien (Jenks and Brake 2001; Sigren et al. 2014). Introduced from Europe in the 1870s, the most widespread dune alien in New Zealand is the dune-builder, Ammophila arenaria or marram. Extensively planted from 1910–1930’s (McKelvey 1999), it had occupied 74% of dunelands by 1992 (Partridge 1992). Also deliberately used to stabilise dunes are a range of coniferous species, especially Pinus radiata (Californian Monterey Pine), which was trialled along the Manawatū coast, south-western North Island, New Zealand, from 1915 (Gadgil and Ede 1998). Widely planted from 1950–1980, it effectively eliminated dune transgressions for the lifetime of the trees. Growth was originally hampered by nitrogen deficiency (McKelvey 1999), so Lupinus arboreus (tree lupin) was introduced from California to ameliorate the deficiency (Gadgil 2002).

New Zealand is rated a global hotspot for invaders (Heywood 1989; Dawson et al. 2017; Tordoni et al. 2021). New Zealand’s vulnerability is a consequence of its recent settlement by both Māori and Europeans, coincident with development of agriculture, increased trans-oceanic traffic and previously lax border controls (Heywood 1989; Asmus and Rapson 2014). Dawson et al. (2017) reported that coasts are especially prone to invasion, and, not surprisingly, many alien dune stabilisers subsequently show themselves as invasive, spreading to other dune areas, and sometimes beyond. Habitat changes wrought by such invaders may also facilitate the invasion of other available aliens. In New Zealand, alien marram traps sand readily, burying the less competitive native sand-binders (Sykes and Wilson 1990; Hilton et al. 2005), and rapidly invading other native vegetation (Esler 1970; Hilton 2006). And of course silviculture, e.g., of pines, completely eliminates native dune vegetation and is an ongoing source of invaders (Peltzer 2018), while tree lupin has become invasive in dunes and waste areas, somewhat slowed by the spread of a fungal blight during the 1980’s (Dick 1994; Gadgil 2002), to which it is now becoming adapted (GLR, pers. obs.). These aliens have the capacity to further stabilise mobile sand, increasing the threat level to mobile dunefields.

New Zealand is an exemplar of coastal invasion processes which, while completed in some parts of the world, are still under way in others. Approximately 10% of New Zealand’s coastline is faced with dunes (Gadgil and Ede 1998). While some dunelands are well documented (Sykes and Wilson 1987, 1991; Roxburgh et al. 1994; Pegman and Rapson 2005; Murphy et al. 2012; Hilton and Konlechner 2022), most are poorly studied. Thus, their vulnerability to over-stabilisation and their exposure to alien invasions is equally poorly understood.

Our focal dunefield is Tawhirihoe (pronounced “Tar-firry-ho-e”) Scientific Reserve, the most important and most mobile dune reserve on the Manawatū coast of the lower North Island. Various methods of sand stabilisation by bioengineering have been explored there for more than a century (Gadgil and Ede 1998). Other studies have focussed on botany (Carnahan 1957), dune dynamics (Esler 1970), particularly blowouts (Hesp 2001, 2002), and the nature and dynamics of dune slack wetlands (Esler 1969; Singers 1997; Yukhnevich 2015; Murphy et al. 2019; Villacís Lozada and Rapson 2021, 2022). Our aim is to assess the impact of alien invaders on dune mobility and on the succession of native vegetation, thus helping shape managerial responses.

In mid-summer of 2009/2010, vegetation was sampled using 52 quadrats, 1 m × 1 m square, placed at random over the dry portions of the mobile dunefield of Tawhirihoe Scientific Reserve. Proportionately appropriate to the mobile duneland, an additional 12 wetland quadrats from those reported in Murphy et al. (2019) were randomly selected to represent the six wetlands.

Within each quadrat, the slope, maximum standing vegetation height, and percentage covers of sand (< 2 mm diameter), stones (≥ 2 mm diameter), wood and dead material, and live vegetation were recorded. Cover of each individual plant species was estimated as the vertical projection of all its foliage onto a horizontal surface, effectively recording its shadow under solar zenith (Rapson 2017). Aspect was also recorded for each quadrat and reported as the cosine of the angle in radians, so values closer to 1 are more northerly (i.e., warmer), while those closer to -1 are southerly. Visual estimation of soil moisture was on a 1 (dry) to 3 (wet) scale. Values of 1 (none) to 5 (extreme) were assigned reflecting the degree of disturbance from wind, rabbits and/or vehicles.

Data on percent covers (square-rooted) of species in each quadrat were clustered using Euclidean distance and Ward linkage metrics in SYSTAT (Systat 1998), to give a strong, though somewhat arbitrary, dendrogram. This was pruned to give seven vegetation types, accepted here as communities. A principal component analysis (PCA) of the untransformed data (to improve the spread of quadrats) was run in Canoco (ter Braak and Smilauer 2002). The community types from the cluster analysis were superimposed on the PCA’s first two axes. For each community type, species’ composition and environmental variables from each quadrat were averaged and tabulated. Community types were mapped from the sampling points (quadrats) overlain on the aerial photograph of 2012 (the only available date close to the field survey), and their area extrapolated using the pixel counter in Adobe Photoshop Co., San Jose, USA.

The importance of invasion by alien plants in the reserve was explored by calculating the relative cover ratio of natives and aliens, by species and by community type, and through fitting a contour plot to the spatial data for alien species’ cover as a proportion of total cover, using linear interpolation following Lodwick and Whittle (1970) in Systat (1998). Changes in vegetative cover over time were inferred from the thesis of Singers (1997), extrapolating from his shoot-presence data and associated environmental measurements in 271 random quadrats of 2 m × 2 m over the mobile dunefield, as well as from personal knowledge (of GLR).

Of the 64 random quadrats, three contained only bare sand and were on flattish sites, highly disturbed by wind and vehicles (Table 1). The remaining 61 quadrats had a mean vegetation cover of 60%. A total of 65 plant species was recorded, with 26 being native. On average, each species was found in 6.5 quadrats, with 22 species found in only one quadrat each. The most frequent species was the alien herb Hypochaeris radicata (found in 52% of quadrats), followed by four alien grasses, Holcus lanatus (46%), Arrhenatherum elatius (38%), Lagurus ovatus (36%) and Ammophila arenaria (33%). The most common native species were the sedge Ficinia nodosa (38%; wiwi) and the sand-binder Spinifex sericeus (25%).

The dendrogram (Table 1) splits quadrats initially into two clusters, the first mostly with low sand cover and high abundance of aliens Ammophila arenaria and Holcus lanatus. The second cluster generally has much higher sand cover and abundance of two natives Apodasmia similis (oioi, pronounced oi’oi’; Restionaceae) and Ficinia nodosa. These clusters were further divided into 3–4 communities each. The first two PCA axes (Principal Components Analysis; Figure 3) explained 22.0 and 12.6% of the variation in the data respectively. The second axis was influenced by native species, while the first responded to the high levels of alien grasses, though the composition of the latter was more variable.

Table 1.

Download as

CSV

XLSX

Mean environmental variables and summary features of the vegetation for each community, and for the quadrats containing only bare sand. For aspect, values closer to +1 are more northerly (i.e., warmer), while -1 is south. Soil moisture is scaled between 1 (dry) and 3 (wet), while disturbance is between 0 (low) and 5 (extreme).

Variate	Bare sand	Spinifex Foredune	Turfed Wetland	Sedge-land	Stone Plain	Oioi Rushland	Marram Duneland	Exotic Grassland
No. of quadrats sampled	3	9	3	15	11	5	13	5
Environment
Distance to coast along slack (m)	291	193	361	539	359	447	357	434
Aspect (Northings)	0.800	0.101	1.000	-0.023	0.333	1.000	0.349	0.228
Angle of greatest slope (°)	3.7	8.8	0.0	6.5	6.7	0.0	9.3	10.4
Soil moisture	1.3	1.0	2.3	1.3	1.2	1.7	1.0	1.3
Wind disturbance	3.7	3.3	2.7	1.6	2.9	3.0	2.3	1.8
Vehicle disturbance	2.3	0.1	0.0	0.2	0.4	0.0	0.3	0.4
Rabbit disturbance	0.0	0.4	0.0	1.2	0.5	0.2	0.3	0.6
Sand cover (%)	100.0	73.1	35.0	9.4	84.7	8.4	47.7	1.0
Stone cover (%)	0.0	0.0	11.7	0.0	5.7	0.0	0.3	0.0
Wood cover (%)	0.2	0.1	0.0	1.8	0.4	0.0	0.5	3.0
Rabbit pellet cover (%)	0.0	0.1	0.0	0.1	0.1	0.0	0.0	0.0
Vegetation
Vegetation height (cm)	0	48	55	100	66	128	84	131
Cover of live vegetation (%)	0.0	30.0	55.0	84.7	6.5	90.0	55.5	95.0
Cover of dead vegetation (%)	0.2	3.2	2.2	11.4	5.3	5.8	4.8	10.2
Total no. spp. in community	0	22	17	35	28	15	30	13
No. of alien spp. in community	0	16	3	29	19	7	23	10
Max. no. of spp. per quadrat	0	10	11	13	12	8	14	7
Mean no. of spp. per quadrat	0.0	5.8	8.7	8.8	4.1	6.0	9.9	5.8
Mean no. of alien spp. per quadrat	0.0	3.8	2.0	7.5	3.1	2.6	8.1	4.8
Mean cover of natives per quadrat (%)	0.0	26.9	32.7	15.3	1.3	71.0	16.1	5.7
Mean cover of aliens per quadrat (%)	0.0	5.3	24.7	76.3	5.6	20.5	43.3	94.7
Prop. per quadrat of vegetative cover as aliens	0.0	0.17	0.43	0.83	0.81	0.22	0.73	0.94

Figure 3. 

Principal Components Analysis of the sites sampled, with colours and ellipses (inclusive of most samples) identifying the seven communities recognised by the dendrogram.

Changes since the first detailed survey of this area by Singers (1997) include the addition of 11 newly-recorded alien species, of which six are grasses, and the loss of 31 alien species, mainly wasteland weeds. One native species (Tetragonia trigyna, Aizoaceae, a species of shady sands) was added, though 14 were lost, including six species of early successional turf wetlands.

The spatial size of the largest wetland (numbered 1 in Singers 1997; Murphy et al. 2019), was greater in 1996 (Villacís Lozada and Rapson 2021, 2022). When analysed by Singers (1997), the wetland had an intermix of communities, covering the Turfed Wetland and Oioi Rushland types which we also identified. These are interspersed with “Exotic Grassland”, Singers’ version of which has high frequencies of Holcus lanatus, Hypochaeris radicata, Leontodon taraxacoides and the native Ficinia nodosa (Singers 1997). Thus, it somewhat resembles the Sedgeland community we report, though at lower vegetation cover. The other major change in the mobile dunefield is the loss of wetland turf vegetation in the south-eastern corner (Singers 1997), due to ongoing vehicular disturbance and infilling.

In our survey, the most widespread community is now Marram Duneland at 29% coverage of the dunefield, while the remaining communities contribute an average of 8% cover each (Figure 5B). In the five communities with > 55% cover of vegetation, alien species dominate in three (Figure 6A), which, broadly-speaking, occur in the rear portion of the reserve (Figure 6B). The cover-rich exceptions are Turfed Wetland and its later successional stage, Oioi Rushland, which are more centrally located, though even within these communities, about 20% of the vegetative cover is alien. Stone Plain is also dominated by aliens, but occupies only 6% of the mobile dunefield (Figure 5B). The greatest cover of natives (83%) is in Spinifex Foredune, which dominates the sea-wards edge of the dunefield (Figure 5A).

Overall, 21% of the studied area is still in mobile sand (Figure 5B), along with bare sand also exposed within communities. Averaging the communities by their contents and extrapolating proportionately across the dunefield, bare sand occupies another 33% of the visible surface. Increased cover of aliens appears associated with a decrease in the coverage of mobile sand (Figure 7A), for which changes over time show a loss rate of approximately 1% per annum (Figure 7B).

Figure 6. 

A) Cover of native (green, above the x axis) and alien (red, below) species in each community. B) Contour plot of the spatial distribution within the mobile dunefield of the proportion of the vegetation cover which is comprised of alien species. Red = more alien, green = more native, blue = plage.

Figure 7. 

Relationships between A) covers of aliens and bare sand in each community, and B) proportional change of bare sand in the mobile dunefield over time. For B) the first and last values are based on literature and personal observations of GLR.

In New Zealand normal dune successions trend towards forested vegetation (Smith et al. 1985; Sykes and Wilson 1991; c.f. Pegman and Rapson 2005). But at Tawhirihoe Scientific Reserve only the second axis of our PCA depicts the natural succession, while the first axis is instead dominated by habitat types with a high cover and diversity of aliens (Figure 3), particularly alien grasses, such as marram.

The native / marram balance on the foredunes

At Mason Bay, Stewart Island, in southern New Zealand, marram’s invasiveness is demonstrated by its’ covering 75 ha in 40 years (Hilton et al. 2005), and the need for a bidecadal control programme to restore pristine-looking foredunes of pingao (Ficinia spiralis; Hilton and Konlechner 2022). Along the Manawatū coast there are occasional patches of pingao, some of which have been planted by iwi (local representatives of the indigenous people), who value the foliage for weaving. But pingao is not particularly common where sands are more mobile (Esler 1970), as Spinifex builds dunes faster. Marram builds foredunes even faster than Spinifex, though it is less salt-tolerant (Sykes and Wilson 1988). However, marram dunes eventually become over-steepened, and destabilise (Figure 8A).

Figure 8. 

Vegetation change over time at Tawhirihoe Reserve. A) Marram turret being undercut by an inland movement of the Rangitikei River mouth; note four juvenile Homo sapiens for scale; March 2007. B) Restoration planting of Spinifex to form a new foredune where movement of the Rangitkei River mouth has exposed fresh sand in front of an earlier restoration planting of a marram foredune; August 2008. C) The site of Wetland 1 inundated in a wet winter, backed by Oioi Rushland, and surrounded by dunes vegetated with alien shrubs, including Lupinus arboreus in the foreground and Acacia sophorae in the background; October 2022. D) Looking across Wetland 2 towards Wetland 6 in an area formerly in Oioi Rushland, Marram Duneland and Stone Plain, showing aliens to ≈ 70% cover; October 2022. Images: Jill Rapson.

The Spinifex Foredune is the most natural community at Tawhirihoe, and despite low levels of invasion by alien dune specialists (e.g., Carpobrotus edulis) and hardy generalists (e.g., Hypochaeris radicata), is probably similar to that occurring before European settlement (Cockayne 1909, p. 15; Johnson 1992; Partridge 1992; Pegman and Rapson 2005). Spinifex’s long, downward-flowing rhizomes root readily into fresh sand (Cockayne 1909), repairing erosion of, and undercutting of, the foredune, such as occurs during tidal or storm surges, so that it can out-compete marram on active foredunes (Esler 1970). Since the Manawatū coast is prograding (Hesp 2001; Villacís Lozada and Rapson 2022), management is now required to vegetate the foredunes with Spinifex (Figure 8B) to counteract the earlier plantings of marram (Cockayne 1909; p. 12; McKelvey 1999).

In New Zealand problems with alien weeds are immense, especially on conservation lands (Williams and West 2000; Howell 2008). At Tawhirihoe ≈ 70% of the reserve is planted in alien pine to “protect” farmland in the hinterland from dune mobility. The remainder of the reserve, the mobile dunefield, has 31% coverage of aliens, from 39 species. Of the 31 alien species recorded by Singers (1997), all still occur in the Manawatū, and most are probably still present within the reserve, so that their absence from the mobile dunefield probably reflects small differences in habitat availability or sampling, or random turnover of invaders. With 11 new species recorded in this survey (unlikely to have been missed by Singers (1997)), these alien-dominated communities seem to be accelerating the invasion of other aliens.

A portion of these alien species inevitably becomes extremely aggressive, influencing the ecology of the mobile dunefield. The nearby Rangitikei River estuary was the invasion site of European horsetail, Equisetum arvense, introduced by a local herbalist about 40 years ago. Resistant to all control efforts there (Esler 1978; Horizons 2013), horsetail subsequently spread into Tawhirihoe Reserve. Randomly unsampled in this survey, it is present in Exotic Grassland, and currently forms aggressive swards in disturbed drainage ditches around the periphery of the reserve. Another major weed is Senecio glastifolius, a South African asterad of fynbos, locally called pink ragwort (CABI 2017). It forms thickets in the rear dunes, with semi-woody tufts to 2 m tall, and has a canopy of bright pink capitula in mid spring, with vast seed production and pappus-based dispersal (Beautrais 2013; CABI 2017). Very invasive, several years of management at Tawhirihoe have failed to keep it in check, due to constant reinvasions from coastal populations situated to the north-west across the Taranaki Bight. In 2006 two individuals of Osmunda regalis (royal fern) were found amongst Oioi Rushland in the centre of Wetland 1, around a falcon “nest” (a terrestrial scrape; Falco novaeseelandiae). First recorded from the nearby Rangitikei region (Field 1890), royal fern is considered a conservation threat to wetlands in New Zealand (Howell 2008), as it displaces smaller native species (Martin 2002). At Tawhirihoe the plants were removed and vouchered (MPN 42111), and no reoccurrence has been noted. Also present at Tawhirihoe, though not sampled, Acacia sophorae (sand or golden acacia; from Australia’s east coast) is a very invasive shrub, forming dense thickets 3–6 m tall, offering no habitat for native species. Herbicided in the past, the remaining plants continue to pose a threat, but hope is held for biocontrol using a bud-galling wasp (EPA 2022). Tawhirihoe Reserve should be monitored for such weeds, and any infestations immediately managed.

Two of Tawhirihoe’s communities dominant on Axis 1 of the ordination are essentially alien (Figure 3). Such communities can cause reduction of native biodiversity and abundance, and loss of rare native species due to over-stabilisation (Acosta et al. 2009; Verhoeven et al. 2014; Tordoni et al. 2021).

Though sometimes mis-directed (Cooper and Jackson 2020), deliberate disturbance to increase dune mobility can be considered a viable management technique (Bird et al. 2020). At Tawhirihoe, the presence of 1.6 km of off-road vehicle tracks in the mobile dunefield (in addition to numerous tracks throughout the pine forest) shows the extent of on-going human disturbance (Figure 5A). Many attempts have been made to protect the native vegetation, but informational signage is largely uninfluential, while excluding vehicles using fencing works only temporarily; even when wires do not get cut, fences are quickly buried by mobile sand. New vehicle routes are formed continually, though in areas of mobile sand, tracks also dissolve rapidly, except for now unreachable remnants which may persist as landscape features for decades. Contrarily, intermittent traffic helps to keep the dunelands open, as the associated compression and disturbance kills larger plants, especially aliens, and lowers the sand surface closer to the water table, providing habitat for rare turf species. For example, when the traffic route through the middle of the south-eastern-most wetland noted by Singers (1997) became inaccessible, the wetland dried out and lupin scrub invaded. But as a management tool, traffic is too unpredictable and illlimitable to be effective.

Bird et al. (2020) recommended grazing as a tool to disturb dunes. But no mammalian grazers are indigenous to Tawhirihoe’s dunes. Instead, alien deer (red and sambar) are increasing in the duneland and surrounding plantations, due to reduced hunting during the Covid pandemic, but also in response to the increasing availability of alien herbage which dominates their sand-country diet (> 90% by weight; Stafford 1997). Since their introduction in the 1850’s (King 2017), rabbits have always been common on New Zealand’s dunes, and while persistent herbivores of aliens, they do more harm than good, probably having a role in the loss, perhaps extinction, of the subshrub Pimelea actea (Rapson 2018a).

Though counter-intuitive, destabilising the sand so that it moves inland (Hesp 2002), seems the most cost-effective option for retention of rarer natives. Breaching the foredunes, perhaps by herbiciding Spinifex would successfully remobilise some sand, and allow the natural formation of dune slacks and perhaps their associated wetlands. But the most likely result inland of the foredunes would be reinvasion of aliens, unless ongoing selective weeding occurs to favour natives.