Content uploaded by Warren M. King
Author content
All content in this area was uploaded by Warren M. King on Mar 08, 2023
Content may be subject to copyright.
- Properties of ecotones: Evidence from five coastal ecotones - 579
Journal of Vegetation Science 14: 579-590, 2003
© IAVS; Opulus Press Uppsala.
Abstract. Several properties have been suggested to be char-
acteristic of ecotones, but their prevalence has rarely been
tested. We sampled five ecotones to seek evidence on seven
generalizations that are commonly made about ecotones:
vegetational sharpness, physiognomic change, occurrence of a
spatial community mosaic, many exotic species, ecotonal
species, spatial mass effect, and species richness higher or
lower than either side of the ecotone.
The ecotones were in a sequence from scattered man-
groves, through salt marsh, rush-marsh, scrub, woodland, to
pasture. We developed a method to objectively define, by
rapid vegetational change, the position and depth of an
ecotone, identifying five ecotones. Their positions were con-
sistent across three sampling schemes and two spatial grain
sizes. One ecotone is a switch ecotone, produced by positive
feedback between community and environment. Another is
anthropogenic, due to clearing for agriculture. Two others
are probably environmental in cause, and one may be largely
a relict environmental ecotone.
Sharp changes in species composition occurred. Three
ecotones were associated with a change in plant physiognomy.
In two, the ecotone was located just outside a woodland
canopy, in the zone influenced by the canopy. Community
mosaicity was evident at only one ecotone. There were few
strictly ecotonal species; many species occurred more fre-
quently within ecotones than in adjacent vegetation, but there
were never significantly more ecotonal species than expected
at random. There was little evidence for the spatial mass effect
reducing ecotonal sharpness, or leading to higher species
richness within ecotones. Species richness was higher than in
the adjacent habitat in only one ecotone.
It seems that supposedly characteristic ecotone features
depend on the particular ecological situation, and the ecology of
the species present, rather than being intrinsic properties of
ecotones.
Keywords: Boundary; Ecotonal species; Edge; Exotic spe-
cies; Mangrove; Marsh; Mosaicity; Spatial mass effect; Spe-
cies diversity; Salt marsh.
Nomenclature: Connor & Edgar (1987) and references therein,
and Stace (1997), except where indicated.
Introduction
Ecotones – abrupt edges in vegetation – have long
fascinated ecologists, and challenged them to find causes
and resulting patterns. The existence of ecotones has
been used to test between competing ecological theories,
since some theories predict that most boundaries between
communities will be abrupt (Odum 1983; Gilpin 1994),
and others predict that abrupt boundaries will occur mainly
when there are abrupt environmental changes (Wilson &
Agnew 1992; Auerbach & Shmida 1993). Ecotones are a
basic unit in landscape studies (Hansen et al. 1988; Wiens
1992), and may have significance for management in the
face of climate change, since such effects are likely to be
seen first at ecotones (Neilson 1991; Kupfer & Cairns
1996; Allen & Breshears 1998).
The term ‘ecotone’ was introduced by Clements
(1905) as “The line that connects the points of accumu-
lated or abrupt change in the symmetry. …Ecotones are
well marked between formations, particularly when the
medium changes: they are less distinct within forma-
tions”. In spite of suggestions that, for etymological
reasons, ‘ecotone’ should have implications of stress
(van der Maarel 1990), most authors have followed
Clements in defining it in terms of spatially-rapid vege-
tation change (e.g. Tansley & Chip 1926; Odum 1983;
Gosz 1993). We use such a definition here: “A zone
where directional spatial change in vegetation (i.e. quali-
tative and quantitative species composition) is more
rapid than on either side of the zone” (Lloyd et al. 2000).
We here develop a method for locating and delimiting
ecotones, based on ordination scores.
Several features have been proposed as characteris-
tic of ecotones, but there is little documentation. In this
paper, we search for evidence of one defining feature of
ecotones: a. the sharpness of vegetational transition, and
six hypotheses (‘b’ to ‘g’):
Properties of ecotones: Evidence from five ecotones objectively
determined from a coastal vegetation gradient
Walker, Susan1,2; Wilson, J. Bastow1*; Steel, John B.1; Rapson, G.L.3;
Smith, Benjamin1,4; King, Warren McG.1,5 & Cottam, Yvette H.6
1Botany Department, University of Otago, P.O. Box 56, Dunedin, New Zealand; 2Present address: Landcare Research, Private
Bag 1930, Dunedin, N.Z.; 3Ecology Group, Institute of Natural Resources, Massey University, Palmerston North, N.Z.;
4Present address: Climate Impact Group, Plant Ecology, Department of Ecology, Ekologihuset, Lund University,
SE-223 62 Lund, Sweden; 5Present address: CRC for Weed Management Systems, New South Wales Agriculture, Orange
Agricultural Institute, Forest Road, Orange, NSW 2800, Australia; 6Institute of Food, Nutrition and Human Health,
Massey University, Palmerston North, N.Z.; *Corresponding author; Fax +6434797583; E-mail bastow@otago.ac.nz
580 Walker, S. et al.
b. change in physiognomy (especially plant height);
c. community mosaicity (i.e. the occurrence of a spatial
vegetational mosaic);
d. ecotonal species (i.e. the existence of species re-
stricted to ecotones);
e. spatial mass effect (i.e. continued migration of
propagules into an ecotone, and establishment of
plants, of species not able to form self-sustaining
populations there);
f. species richness (suggested to be higher or lower in
ecotones than in the communities to either side); and
g. exotic species (being particularly common in eco-
tones).
We use vegetation data collected at a coastal site in
northern New Zealand, that seemed to show discon-
tinuities in the vegetation, and proved to contain a
number of ecotones.
Methods
Sampling
We sampled a sequence of vegetation at Emauha
Point in Northland, New Zealand (172∞52' E, 34∞33' S;
mean annual temperature 15.8 ∞C; rainfall 1276 mm;
precipitation deficit 106 mm; Leathwick et al. 2002).
The sequence went from mangrove, through a rush
community (primarily Juncus krausii var. australiensis,
marsh with shrubby trees of Leptospermum scoparium
(manuka), woodland of L. scoparium and Kunzea
ericoides (kanuka), to pasture. Because this is a tidal
region, we expected that some of these would be envi-
ronmental ecotones, i.e. ecotones caused by a sharp
environmental change. The Kunzea to pasture ecotone
is clearly anthropogenic, caused by clearing of the wood-
land for pasture.
All vegetation sampling was in 0.5 m ¥ 0.5 m quadrats,
recording the shoot presence of all vascular species.
Seedlings were recorded separately, a seedling being
defined as: (1) for a woody species: unbranched and less
that 10 cm tall, or (2) for a grass or rosette herb: with
three or fewer leaves, or (3) for other forbs: with five or
fewer leaves. The height of the vegetation, and, where
appropriate, the girth of the largest woody stem in each
quadrat, were recorded.
There were three parts of the sequence from man-
grove to pasture in which vegetation change seemed to
be more rapid (Fig. 1a): a. mangrove to rush, b. rush to
woodland, and c. woodland to pasture.
In each of these three parts, an intensively-sampled
transect (Fig. 2a) was randomly located within a 150 m
wide sector of the sequence. The transect was at right
angles to the apparent ecotone, and extended into the
apparently more uniform vegetation on either side. For the
length of each intensive transect, sampling was in eight
quadrats placed contiguously across the transect, and
contiguously along it. Between th e three intensive transects,
Fig. 1. (a) Profile through the
Emauha Point study area from
measured elevations and canopy
heights, and (b) Ordination
scores along the entire transect
from the equally-weighted ordi-
nation, and the rate of change in
ordination score with distance
(i.e. the slope of the regression
of ordination score on distance,
i.e. its first derivative) within a
window width of 25 m. Vertical
arrows indicate the positions of
the centre of the five ecotones
(I to V) as identified by a peak in
the rate of change.
- Properties of ecotones: Evidence from five coastal ecotones - 581
lines of non-contiguous quadrats were placed at 5-m
intervals. In total, 4092 quadrats were sampled. Trunk
sections were taken from five of the largest trees at two
points in the sequence where rapid physiognomic
change was apparent, and their age estimated by ring
counts. Elevations were recorded at each quadrat in
each intensive transect using surveying equipment.
The height of high tide was similarly recorded on seven
days, and used to calibrate the elevations to mean tide
level. Light intensity available to seedlings was meas-
ured at 50-cm height on a 1-m grid over the intensive
transects (except in the mangrove-to-rush transect,
where there was no tall vegetation) using a Licor
quantum PAR sensor, and expressed as a percentage of
the incident light level in the open, taken every four
quadrat readings. In the mangrove-to-rush transect, the
percentage of bare mud was estimated visually.
Tests of ecotone features
Analyses were performed at the sampled spatial
grain of 0.5 m ¥ 0.5 m, and, for the intensive transects,
also by lumping square blocks of adjacent quadrats at
grains of 1 m ¥ 1 m, 1.5 m ¥ 1.5 m and 2 m ¥ 2 m.
a. Sharpness of vegetational transition (location and
width of ecotones); b. Change in physiognomy:
The exact position of ecotones and their widths were
determined from ordination scores. Three sets of
ordinations were performed using Detrended Corre-
spondence Analysis (Hill & Gauch 1980) of the species
presence/absence data (excluding seedlings):
1. Overall: This used the entire data set of 4092
quadrats of 0.5 m ¥ 0.5 m, and ensured that the areas
selected for intensive sampling were indeed those where
ecotones occurred.
2. Equally-weighted: Here we also used data from
across the entire ecotone sequence, but selected one, of
the eight 0.5 m ¥ 0.5 m quadrats across a row diagonal to
the transect, at random in each 5 m distance interval in
each intensive transect. This was to check for the effect
of differences in sampling intensity.
3. Separate: We performed separate ordinations for
each intensive transect using all the quadrats from that
transect. These were repeated at two different spatial
grain sizes, 0.5 m ¥ 0.5 m and 1 m ¥ 1 m, because it has
been suggested that vegetation spatial grain might be
important in identifying changes at ecotones (Hansen et
al. 1992). Coarser grain sizes than 1 m ¥ 1 m were not
analysed because the number of quadrats would have
been too small.
To verify that the first ordination axis represented
the sequence along the transect, we calculated the rank
correlation of ordination score with distance along the
sequence from mangrove to pasture.
To locate the exact position of ecotones, a moving-
window regression method was developed, using the
ordination scores, similar to those previously used to
analyse two-phase patterns of vegetation (Whittaker &
Naveh 1979; Whittaker et al. 1979). We represented
species composition with Axis 1 ordination scores (aver-
aged over the eight quadrats in a row within intensive
transects), and calculated the rate of change in ordination
Fig. 2. (a) Detail of layout of an intensively-sampled transect,
(b) Demonstration of method of calculating the rate of change
in species composition using a moving window, and (c) as an
example, the definition of the positions of Ecotones IV and V
from the second derivative of the ordination scores.
582 Walker, S. et al.
score with distance, as the linear least-squares regres-
sion slope within a window of n (see below) ordination
scores along the vegetation sequence (Fig. 2b). The
midpoint of an ecotone was identified as the position of
maximum rate-of-change in ordination scores along the
length of the transect. Therefore, it is seen as a peak in
the regression slope (Fig. 2c; rate of change). To deter-
mine the width of an ecotone, a second moving-window
regression was performed of the slopes (i.e. rates of
change) produced by the first regression on distance.
Points of inflection in the scores of the first regression
(i.e. the second derivatives), are seen as maxima and
minima in the scores of this second regression, and are
interpreted as the boundaries of the ecotones (Fig. 2c;
second derivative). These boundaries were used to di-
vide the vegetation sequence into ecotones and non-
ecotone zones.
In a moving-window regression method, a small
window contains too much noise, whereas a wide win-
dow is liable to smooth over the vegetation change of
interest. We therefore chose optimum window widths
after performing preliminary analyses at a range of
widths. A 25 m window was used for the Overall and
Equally-weighted ordinations, and a 6 m window was
used for the Separate ordinations.
c. Community mosaicity:
We developed indices of community mosaicity to
test whether a small-scale spatial mosaic is characteris-
tic of ecotones (Pound & Clements 1900; Gosz 1992;
Neilson et al. 1992). We applied these to each row of
eight contiguous 0.5 m ¥ 0.5 m quadrats (i.e. a row
transverse to the direction of the transect; Fig. 3) within
each intensive transect. First, an ‘adjacent’ mosaicity
index, MA, was calculated by examining the spatial
turnover in species occurrences (excluding seedlings)
between adjacent pairs of quadrats (Fig. 3):
MN
n
A
P
s
S
=-
=
Â1
1
(1)
where NP = the tally, across the eight quadrats in a row
of the transect, of pairwise differences, i.e. cases where
the species occurs in one or other but not in both quadrats
of the pair (X- or -X) (Fig. 3); n = the number of quadrats
in a row (here n = 8, therefore there are n – 1 = 7 adjacent
pairs); and S = number of species.
A drawback of MA is that if the vegetation zonation
is diagonal to the transect, some zonation would appear
as mosaicity. Therefore, a second index, the ‘trios’
mosaicity index, MT, was calculated from trios of
quadrats (Fig. 3):
MN
n
T
T
s
S
=-
=
Â
2
1
(2)
where: NT = the tally, across the eight quadrats in a row
of the transect, of trio reversals, i.e. instances in which a
species was present in both the first and third quadrats of
a trio but not in the second (X-X), plus the number of
instances in which a species was present in the second
quadrat but not in the first or third quadrat (-X-) (Fig. 3).
Indices MA and MT were calculated using the 0.5 m ¥ 0.5
m and the 1 m ¥ 1 m quadrats. A high value for either
index would indicate a community mosaic.
d. Ecotonal species:
For each intensive transect, possible ecotonal species
were noted as those (excluding seedlings) which occu-
pied a higher proportion of quadrats in the ecotones than
in the zones on either side. The expected proportion of
quadrats occupied by a species in each zone was esti-
mated as the average across the whole intensive transect.
However some species would be expected to have their
highest frequency in an ecotone, if they were distributed
at random. Therefore, a null model was constructed in
which a length of transect equal to the observed ecotone
was placed at random along the transect (or in transects
with two ecotones, the positions of both were randomized,
with the restriction that they could not overlap), and the
number of ‘ecotonal’ species was calculated. Signifi-
cance was calculated as the proportion of 2000
randomizations in which the number of ecotonal species
was equal to, or more extreme than, the number ob-
served, multiplied by 2 to give a 2-tailed test.
e. Spatial mass effect:
For each species for which seedlings were recorded,
the data were examined for the occurrence of seedlings
beyond the range of adult plants, which would be evi-
dence for the continued migration of propagules and
establishment of plants in areas where the species was
not able to form a self-sustaining population.
f. Species richness; g. Exotic species:
In each of the intensive transects, total species
richness per quadrat, and the proportion of exotic
species, were compared between each zone (ecotones,
above and below ecotones), using Analysis of Vari-
ance with unequal replication, testing against within-
zone variance.
- Properties of ecotones: Evidence from five coastal ecotones - 583
Results
a. Sharpness of vegetational transition (location and
width of the ecotones)
The first ordination axis represented well the se-
quence along the transect, with distance and ordination
score having a rank correlation of 0.95. An ecotone
(zone of rapid change in the vegetation) is shown by a
high rate of change in ordination score (Fig. 2c). By this
criterion, the Equally-weighted ordination showed five
main ecotones over the area (Ecotones I to V; Fig. 1), all
located within the three intensive transects:
Mangrove to rush transect:
Mangrove/Rush (Ecotone I)
Rush to woodland transect:
Rush/Marsh (Ecotone II)
Marsh/Woodland (Ecotone III)
Woodland to pasture transect:
Woodland/Woodland-edge (Ecotone IV)
Woodland-edge/Pasture (Ecotone V)
The three ecotones apparent in the Equally-weighted
ordination at the uphill woodland edge (Fig. 1b) re-
solved into two in the finer scale analysis of that
intensive transect (Fig. 2c). Otherwise, the positions of
these ecotones, and the vegetation gradients identified,
were almost identical in the Overall and Separate
ordinations, and using spatial grains of 1 m ¥ 1 m and
0.5 m ¥ 0.5 m. Thus, the ecotone-location method seems
to be robust against some variation in sampling intensity
and spatial grain size. We use the most fine-grained
information for further work.
Ecotone I: Mangrove/Rush:
A sparse and dwarf monoculture of the mangrove
Avicennia marina at the lowest elevation was replaced at
Ecotone I (16.5 m) by a community dominated by Juncus
krausii, with sporadic occurrence of salt marsh species
such as Sarcocornia (
∫
Salicornia) quinqueflora (Figs.
1a, 4a). The ecotone occurred at a 0.3 m high erosion
clifflet, convoluted into a mosaic of creeks and islets.
Ecotone II: Rush/Marsh:
At the lower end of this ecotone (230 m), very uni-
form Juncus krausii (rush) with sporadic occurrence of
other salt tolerant species, was replaced at a rise in eleva-
tion, by marsh dominated by the upright-stem cyperad
Baumea juncea and the exotic Gladiolus undulatus, with
scattered Leptospermum scoparium shrubs (Fig. 4b).
Fig. 3. Calculation of Ma (adjacent) and Mt (trio) mosaicity for
a single-species community.
584 Walker, S. et al.
Ecotone III: Marsh/Woodland:
Two maxima in the rate-of-change in species com-
position between 326 and 340 m indicate that Ecotone
III comprised two transitions in rapid succession.
Sarcocornia quinqueflora was not found above this
ecotone, and only scattered Juncus krausii continued
beyond the ecotone. Tall Leptospermum scoparium trees
(those examined being 7, 11 and 15 yr old respectively)
were recorded, for the first time along the vegetation
sequence, just beyond Ecotone III. Ordination scores
fluctuated in the woodland zone between 340 m and 533
m (especially in the ca. 70 m before Ecotone IV; Fig.
1b), with a gradual change to dominance by tall trees of
L. scoparium and Kunzea ericoides, and shrubs of
Coprosma parviflora, over a herbaceous layer (Fig. 4b).
Ecotone IV: Woodland/Woodland-edge:
Continuous woodland of Leptospermum scoparium
and Kunzea ericoides extended across Ecotone IV at
533 - 540 m (Figs. 1a, 4c). However, understorey herbs
which occurred in the marsh (especially rushes and
cyperads) were largely replaced by species more typical
of the pasture.
Ecotone V: Woodland-edge/Pasture:
The beginning of Ecotone V, at 546.5 m, was a
transition from woodland with Kunzea ericoides trees
up to 35 years old to a herbaceous pasture community
dominated by exotic species: grasses Anthoxanthum
odoratum, Axonopus affinis and Vulpia bromoides, forbs
Hypochaeris radicata and Lotus subbiflorus, and the
rush Juncus dichotomus (Figs. 1a, 4c). At the inland
edge of the ecotone (553.0 m) exotic species became
dominant, with the grasses Pennisetum clandestinum
and Poa trivialis prominent.
b. Change in physiognomy
Vegetation canopy height doubled across Ecotone I
(Mangrove/Rush: Fig. 4a), but showed no change in
Ecotone II (Rush/Marsh). Trees of Leptospermum
scoparium replaced shrubs a few metres above the end
of Ecotone III (Marsh/Woodland; Fig. 4b). Canopy
height did not change with the understorey transition of
Ecotone IV (Woodland/Woodland-edge; Fig. 4c), but it
decreased dramatically at Ecotone V as woodland gave
way to pasture. The girths of Leptospermum scoparium
and Kunzea ericoides trees were greatest immediately
before Ecotone V, at the woodland edge (Fig. 4c).
c. Community mosaicity
At Ecotone I (Mangrove/Rush), average values of
mosaicity indices MA and MT at the 0.5 m ¥ 0.5 m grain
size were significantly greater in Ecotone I than in the
communities at each side of the ecotone (P < 0.001 for
both indices; Fig. 5a). At the coarser (1 m ¥ 1 m) grain,
community mosaicity was highest in the rush commu-
nity above the ecotone. There was a general trend for
community mosaicity to increase upshore across
Ecotones II (Rush/Marsh) and III (Marsh/Woodland).
MT (at both grain sizes) and MA (at 0.5 m ¥ 0.5 m)
Fig. 4. Ordination scores and environment characteristics of
the five ecotones: a. Mangrove/Rush (Ecotone I); b. Rush/
Marsh (Ecotone II) and Marsh/Woodland (Ecotone III); c.
Woodland/Woodland-edge (Ecotone IV) and Woodland-edge/
Pasture (Ecotone V).
- Properties of ecotones: Evidence from five coastal ecotones - 585
reached local, though non-significant, maxima within
Ecotone III (Fig. 5a). Neither of the two woodland
ecotones (IV and V) was significantly different in com-
munity mosaicity from their adjacent communities.
d. Ecotonal species
In no ecotone was the number of species reaching
their maximum in the ecotone greater than that expected
at random (Table 1, 2). No species were found signifi-
cantly more frequently in Ecotone I (Mangrove/Rush)
than in either of the adjacent communities. All of the
adult plants of Anagallis arvensis (exotic) and Plagian-
thus divaricatus recorded occurred within Ecotone II
(Rush/Marsh; 10.7 and 14.7 ¥ the number expected at
random). Sonchus oleraceus was found more frequently
within Ecotone II than in the marsh beyond (6.9 times
the frequency of occurrences expected at random). In
Ecotone III (Marsh/Woodland), Briza minor occurred
9.1 times more frequently than expected at random.
Other species considerably more abundant in Ecotone
III than expected at random included Holcus lanatus
(5.2¥), Isolepis reticularus (4.8¥), Juncus articulatus
(5.2¥), J. effusus (6.2¥), J. planifolius (4.5¥),
Lachnagrostis filiformis (4.5¥), Paspalum dilatatum
(4.5¥), and Pennisetum clandestinum (5.5¥). There were
no true ecotonal species in Ecotone IV (Woodland/
Fig. 5. a. Vegetation mosaicity, estimated by two indices
(‘Adjacent’ index MA and ‘Trio’ index MT) at two spatial
scales (dimensions shown in m) in five ecotones (I to V) and in
the adjacent vegetation; b. Species richness in the same zones,
at four spatial scales; c. Proportion of species that are exotic, at
the same four scales.
Table 1. Possible ecotone species for the five ecotones (I to
V), as species more frequent in the ecotone than in the commu-
nity above or below it, excluding cases where the ecotonal
species occurs only one or two times. Entries are the percent
frequency in each zone. Asterisks denote exotic species.
Below Ecotone Above
Ecotone I: Mangrove/Rush
No ecotonal species
Ecotone II: Rush/Marsh
Anagallis arvensis * 0.0 2.9 0.1
Baumea juncea 5.7 68.4 52.2
Lotus pedunculatus * 0.0 16.2 11.7
Plagianthus divaricatus 0.0 9.6 0.0
Sonchus oleraceus * 0.6 5.9 0.7
Ecotone III: Marsh/Woodland
Baumea juncea 52.5 85.8 84.7
Briza minor * 0.0 3.0 0.0
Cotula coronopifolia 1.3 3.0 0.0
Holcus lanatus * 0.1 3.4 1.5
Isolepis reticularis 4.5 43.1 11.9
Isolepis sepulcralis 12.8 37.1 19.5
Juncus articulatus * 0.4 2.6 0.2
Juncus effusus * 0.4 4.3 0.2
Juncus pallidus 2.8 36.6 28.6
Juncus planifolius 1.3 9.9 3.6
Lobelia anceps 34.6 50.9 10.6
Lachnagrostis filiformis 8.5 47.4 3.2
Lolium perenne * 0.1 3.9 2.5
Pennisetum clandestinum * 0.6 8.2 1.7
Paspalum dilatatum * 0.3 78.4 49.2
Ecotone IV: Woodland/Woodland-edge
Anthoxanthum odoratum * 83.6 91.2 59.1
Carex breviculmis 0.8 3.8 0.0
Cirsium vulgare * 0.0 5.0 0.0
Crepis capillaris * 0.8 3.8 0.0
Dactylis glomerata * 0.8 17.5 1.7
Geranium molle * 3.1 5.0 0.0
Holcus lanatus * 13.3 85.0 53.4
Hypochaeris radicata * 75.0 91.2 59.1
Juncus gregiflorus 3.1 5.0 0.0
Lotus pedunculatus * 1.6 87.5 62.5
Lolium perenne * 41.4 61.2 0.0
Trifoium repens * 18.8 56.2 0.0
Veronica arvensis * 2.3 6.2 0.0
Ecotone V: Woodland-edge/Pasture
Carex breviculmis 0.0 7.7 0.0
Coprosma parviflora 46.0 65.4 32.5
Dichondra sp. 0.0 9.6 0.0
Juncus planifolius 0.6 4.8 2.5
586 Walker, S. et al.
Woodland-edge), though all of the ten occurrences of
the creeping herb Dichondra sp. found in the study were
within Ecotone V (Woodland-edge/Pasture).
e. Spatial mass effect
The occurrence of seedlings in ecotones when adult
plants were not present would be evidence for the spatial
mass effect. We examined those species of which seed-
lings were found. In the Rush/Marsh ecotone (II), Anagallis
arvensis, Plagianthus divaricatus and Sonchus oleraceus,
which occurred in the ecotone, occurred as just one, two
and three seedlings respectively beyond the range of
adults, all above the ecotone. This is the opposite pattern
from that proposed by theory: i.e. seedlings of ecotonal
species were outside the ecotone. The other species for
which seedlings were found all occurred as adults in the
Marsh/Woodland (III) ecotone, and also in one or both
of the adjacent communities. They therefore offer no
evidence for the spatial mass effect.
f. Species richness
Species richness was low throughout Ecotone I (Man-
grove/Rush), and was not significantly higher than in
adjacent zones at either grain size. Species richness
increased sharply at both Ecotones II (Rush/Marsh) and
III (Marsh/Woodland) at all spatial grain sizes (Fig. 5b),
but decreased again only after Ecotone III. Only at a
grain size of 0.5 m ¥ 0.5 m were differences between
zones significant. Species richness in Ecotones IV
(Woodland/Woodland-edge) and V (Woodland-edge/
Pasture) was not significantly different to that in adja-
cent communities.
g. Exotic species
In no case was the proportion of exotics significantly
higher in an ecotone than in the two adjoining communi-
ties at the grain size of 0.5 ¥ 0.5 m (Fig. 5c).
The exotic proportion in Ecotone II (Rush/Marsh)
was only slightly and non-significantly higher than above
the ecotone at the 0.5 m ¥ 0.5 m grain size, though at
coarser scale (1 m ¥ 1 m) this was significant. The
proportion of exotics in Ecotone III was generally inter-
mediate between the values in the communities above
and below it. The proportion of exotic species (mainly
pasture species), rose across the transect as a whole,
with a tendency to rise in the ecotones IV (Woodland/
Woodland-edge) and V(Woodland-edge/Pasture), and
reached its highest value in the pasture (Fig. 5c).
Discussion
Types of ecotone
Different types of ecotone can be distinguished on
the basis of the cause of the community sharpness
(Lloyd et al. 2000). The main types are (a) environmen-
tal, caused by a sharp environmental change, (b) switch,
caused by a positive feedback between community and
environment, (c) invasion, where there is invasion of a
dominant species along a front, and (d) anthropogenic,
caused directly by humans. The Mangrove/Rush ecotone
(I) is probably a switch ecotone (Wilson & Agnew
1992), i.e., caused by a positive feedback switch. Where
the sward is intact, the muddy substrate is held together
by the plant roots. Once erosion starts at a point, the
plants no longer hold the substrate together, and it is
eroded further, causing more plant death. It is a charac-
teristic of switches that they can produce an ecotone by
sharpening an underlying gradient, and that they can
create a mosaic (Wilson & Agnew 1992). Both effects
are shown here, where there is a clifflet defining the
edges of a mosaic.
The position of the Rush/Marsh ecotone (II) exactly
correlates with an increase in substrate elevation, from
only just above mean high tide level to about 0.2 m
above. The profile suggests that this might be an old
levee. Ecotone II is clearly an environmental ecotone.
There is a small increase in substrate elevation at the
start of Ecotone III (Marsh/Woodland), which probably
represents an old marsh edge, a few decades old to judge
from the tree ages, and the ecotone may be largely
relictual.
As substrate elevation increases steeply, taking the
plants beyond the reach of flooding, the Woodland/
Woodland-edge ecotone (IV) exhibits a similar situa-
tion to that in Ecotone II. However, drainage is poor,
allowing the occurrence of several marsh species. The
ecotone is clearly environmental in cause.
The Woodland-edge/Pasture ecotone (V) is certainly
anthropogenic, caused by felling of woodland up to this
line, and subsequent grazing, probably with over-sow-
ing of pasture species and fertilising. However, it occurs
Table 2. Number of ecotonal species, defined as having greater
frequency in the ecotone than in the communities on either
side: expected under a null model of random ecotone position,
observed, and the significance of the difference. n.s. = P > 0.05.
Ecotone No. of species
Expected Observed P
IMangrove/Marsh 1.9 1 n.s.
II Rush/Marsh 7.3 7 n.s.
III Marsh/Woodland 8.8 17 n.s.
IV Woodland/Woodland-edge 14.7 16 n.s.
VWoodland-edge/Pasture 8.4 5 n.s.
- Properties of ecotones: Evidence from five coastal ecotones - 587
close to the bottom of the slope, where moisture will
accumulate (Fig. 1). This is probably a case of a switch
based on human/stock activity sharpening an environ-
mental gradient, as both humans and stock avoid the
marshy area below (Wilson & King 1995).
Properties of ecotones
In very little ecotone work is the position of the
ecotone determined objectively. Methods do exist for
boundary detection (see Cornelius & Reynolds 1991)
but ecological studies of ecotones have generally used
subjective methods. Our method examines the rate of
change, rather than the difference between two seg-
ments as in the moving-window method of Cornelius
& Reynolds (1991). We suggest this is more appropri-
ate for analysing ecotones, which are zones of rapid
change rather than points of instantaneous change. We
believe our method is also more appropriate for the
purpose of identifying ecotones along a clear environ-
mental gradient than the wombling method of Fortin &
Drapeau (1995), which is intended for two-dimensional
mapping.
The five ecotones that were identified ranged in
width from 4 m to 14 m. Our method gave almost identical
results for ecotone location and width with the two
spatial grains (0.5 m ¥ 0.5 m and 1 m ¥ 1 m), suggesting
that it is not critically dependent on the spatial grain at
which the vegetation is examined (cf. Fortin 1999),
though 5 m is too large. The Mangrove/Rush ecotone (I)
was the narrowest, probably about 0.5 m at any one
point, but with the mosaic covering 4 m. Clements
(1905), originator of the ecotone concept, commented
that: “Ecotones are rarely sharply defined”. However,
our results demonstrate that quite sharp boundaries can
occur, with rapid replacements of suites of species (i.e.
communities).
Gosz (1992) suggested that physiognomy (life form)
will change across an ecotone, and such a change is
explicit in the ecotone definitions of Tansley (1939) and
Dice (1952). At our study site, changes in physiognomy
occurred in three of the five ecotones (Mangrove/Rush;
Marsh/Woodland, and Woodland-edge/Pasture). We
initially chose the study area because of the changes in
physiognomy, and chose the positions of the intensive
transects by a combination of physiognomy and species
composition, but the final positions of the ecotones were
determined, and two extra ecotones were identified, by
the objective regression method. The change in vegeta-
tion height in Ecotone I coincided with the position of
the ecotone, as identified by species composition, but
the physiognomic changes associated with Ecotones III
and V were just outside the ecotones. The latter ecotones
are in the zone just outside the tree canopy, but still
influenced by it. It is often assumed that the intermedi-
ate community is just inside a forest (Schonewald-Cox
& Bayless 1986), but our results indicate that the true
ecotone, in terms of the most rapid change in species
composition, can be just outside the tree canopy.
Towards an ecotone, conditions will become mar-
ginal for some of the species, and species occurrences
will therefore become more sensitive to environmental
heterogeneity, potentially resulting in a small-scale com-
munity mosaic in species composition (Neilson et al.
1992; Gosz 1992). Indeed, it has been recognized from
the origin of the ecotone concept that there was likely to
be “a mosaic, in which the various pieces now stand out
sharply, and are now obscure” (Pound & Clements
1900). Obviously, mosaicity will depend on the spatial
grain at which the system is viewed, but there is very
little evidence in the literature for mosaics at ecotones.
Mosaicity does not seem to be a necessary feature of
ecotones. The only clear case of mosaicity in this study
was at Ecotone I (Mangrove/Rush), where the mosaic
element comes from channels of mud flat with islets of
salt marsh. The mosaicity is not a direct result of the
ecotone; instead both the ecotone and mosaicity are the
results of the switch. The high mosaicity was not appar-
ent at a spatial grain of 1 m ¥ 1 m (Fig. 5a), because
quadrats of this size included both mud flat and salt
marsh. This emphasizes the scale-dependent nature of
mosaics.
The literature contains the concept of ecotonal
species, i.e. species that are restricted to the ecotone, and
are absent from the two communities on each side of the
ecotone (Clements & Shelford 1939; Hansen et al. 1988;
Neilson et al. 1992). Species with such distribution
patterns might occur if: (a) environmental conditions in
ecotones were intermediate between those of the adja-
cent communities and there were species adapted to
these conditions, or (b) if there were unique environ-
mental conditions in the ecotone (Stoutjesdijk & Bark-
man 1992), or (c) because the species needed resources
from both sides of the ecotone (such as climbers at a
forest edge, requiring the light from outside, but the
trees for support: Putz 1984). There have been few
previous reports of ecotonal species, and in most cases
the evidence has been doubtful (e.g. Anderson et al.
1980; Jose et al. 1996). In this study, we found several
species which occurred considerably more frequently
within ecotones than in the communities adjacent to the
ecotone, though only three which were restricted en-
tirely (within the studied area) to an ecotone, including
some common species (e.g. Isolepis reticularis and
Lachnagrostis filiformis in Ecotone III; Table 1). How-
ever, compared with a null model, we found that in no
ecotone were there significantly more ecotonal species
than expected. There are also species characteristic of
588 Walker, S. et al.
more than one ecotone (Table 1): Carex breviculmis,
Holcus lanatus, Juncus planifolius and Lolium perenne.
Of these, C. breviculmis occurred 12¥ in two ecotones
and only once outside an ecotone. These are more con-
vincing as ecotonal species.
The spatial mass effect is the repeated flux of
propagules, from one (‘source’) habitat in which the
species can reproduce and maintain a population, into
another (‘sink’) habitat in which the species cannot
maintain a population, even though plants can grow
(Shmida & Ellner 1984; Pulliam 1988; Stevens 1992). It
is very difficult to find evidence for the spatial mass
effect (Wilson 1990). However, finding seedlings out-
side the range of adults, in a situation where active
invasion is not occurring, is appropriate evidence. If the
ecotones are sink habitats, this would explain the postu-
lated high species richness there. In our study, seedlings
were recorded beyond the ranges of adult plants for only
a few species. All of these species occurred more
commonly within ecotones than in communities adja-
cent to the ecotones. Therefore the seedling distribu-
tion did not lead to higher species richness within the
ecotones, and no spatial mass effect of the type hy-
pothesized is apparent.
Many ecologists have suggested that richness will
be higher in ecotones than in adjacent communities
(e.g. Leopold 1933; Odum 1983; Petts 1990), but van
Leeuwen (1966) and van der Maarel (1976) suggested
there will be environmental fluctuations at ecotones,
making the environment unfavourable for plant growth,
leading to low species richness. There is little empiri-
cal data relevant to this subject. There have been a few
demonstrations of higher ecotonal richness, but none
are very convincing (Shmida & Wilson 1985; Brothers
1993; Wolf 1993; Carter et al. 1994). Some reports
show species richness in the ecotone to be no different
from that in adjacent communities (Harper 1995; Luczaj
& Sadowska 1997). In our study, species richness was
intermediate between the richness of the two adjacent
communities in two ecotones (I and II), but closer to
the higher-richness community. One ecotone (IV)
showed a slight, and non-significant, tendency to be
less species-rich than either adjacent community. In
the other two ecotones, richness was higher than on
either side, though significantly so only in Ecotone III.
It is clear that high species richness is not an intrinsic
feature of ecotones.
Ecotones, both natural and anthropogenic, contain
a relatively high proportion of exotic species (Norton
1992; Brothers & Spingarn 1992; Risser 1995). A few
examples have also been reported of exotic species
invading ecotones (Ewel 1986; Elias 1992; Puyravaud
et al. 1994; Duggin & Gentle 1998). In our study area,
more than half (21/38) of the species identified as
possibly ecotonal were exotic, out of a total flora of
100 species. The proportion of exotic species in our
study area showed a general increase with distance
from the sea, but at a spatial grain of 2 m ¥ 2 m, there
was indication of a higher exotic representation in all
five ecotones than in adjacent communities. The effect
was consistent across spatial grain sizes only in Ecotone
II, and even there significant only at a spatial grain of 1 m
¥ 1 m. Thus, the evidence in our study for exotics being
more frequent in ecotones is weak.
General conclusions
The literature contains many generalizations about
ecotones. Situations can be found that conform to each,
but none of the generalizations was true of all five of the
ecotones that we investigated. Only one ecotone coin-
cided precisely with a change in physiognomy. There
was some evidence for the existence of ecotonal spe-
cies, but little for the spatial mass effect. There was
evidence for mosaicity, higher species richness and a
higher frequency of exotic species in only one ecotone
each, and the effects often varied with spatial grain size.
We conclude that none of the hypothesized ecotone
features that we examined are intrinsic properties of
ecotones. All the effects result from the particular eco-
logical conditions and the properties of the species
present.
Acknowledgements. We thank Neil Dempster of Paua Sta-
tion for allowing access, Trevor and Gail Bullock of Te Paki
Farm Park for accommodation and assistance, Lisa Forester
for information, and the Pioneer Pog ‘n‘ Scroggin Bush Band
for making their staff available.
References
Allen, C.D. & Breshears, D.D. 1998. Drought-induced shift of
a forest-woodland ecotone: rapid landscape response to
climate variation. Proc. Natl. Acad. Sci. U.S.A. 95: 14839-
14842.
Anderson, P.H., Lefor, M.W. & Kennard, W.C. 1980. Forested
wetlands in eastern Connecticut: their transition zones and
delineation. Water Res. Bull. 16: 248-255.
Auerbach, M. & Shmida, A. 1993. Vegetation change along
an altitudinal gradient on Mt Hermon, Israel – no evidence
for discrete communities. J. Ecol. 81: 25-33.
Brothers, T.S. 1993. Fragmentation and edge effects in central
Indiana old-growth forests. Nat. Areas J. 13: 268-275.
Brothers, T.S. & Spingarn, A. 1992. Forest fragmentation and
alien plant invasion of Central Indiana old-growth forests.
Conserv. Biol. 6: 91-100.
Carter, V., Gammon, P.T. & Garrett, M.K. 1994. Ecotone
dynamics and boundary determination in the Great Dis-
- Properties of ecotones: Evidence from five coastal ecotones - 589
mal Swamp. Ecol. Appl. 4: 189-203.
Clements, F.E. 1905. Research methods in ecology. Univer-
sity Publishing Company, Lincoln, NE, US.
Clements, F.E. & Shelford, V.E. 1939. Bio-ecology. Wiley,
New York, NY, US.
Connor, H.E. & Edgar, E. 1987. Name changes in the indig-
enous New Zealand Flora 1960-1986 and Nomina Nova
IV 1983-1986. N. Z. J. Bot. 25: 115-170.
Cornelius, J.M. & Reynolds, J.F. 1991. On determining the
statistical significance of discontinuities within ordered
ecological data. Ecology 72: 2057-2070.
Dice, L.R. 1952. Natural communities. University of Michi-
gan Press, Ann Arbor, MI, US.
Duggin, J.A. & Gentle, C.B. 1998. Experimental evidence on
the importance of disturbance intensity for invasion of
Lantana camara L. in dry rainforest-open forest ecotones
in north-eastern NSW, Australia. Forest Ecol. Manage.
109: 279-292.
Elias, P. 1992. Vertical structure, biomass allocation and size
inequality in an ecotonal community of an invasive annual
(Impatiens parviflora DC) on a clearing in SW Slovakia.
Ekologia (CSFR) 11: 299-313.
Ewel, J.J. 1986. Invasibility: lessons from South Florida. In:
Mooney, H.A. & Drake, J.A. (eds.) Ecology of biological
invasions of North America and Hawaii, pp. 214-230.
Springer-Verlag, New York, NY, US.
Fortin, M.J. 1999. Effects of quadrat size and data measure-
ment on the detection of boundaries. J. Veg. Sci. 10: 43-50.
Fortin, M.J. & Drapeau, P. 1995. Delineation of ecological
boundaries: comparison of approaches and significance
tests. Oikos 72: 323-332.
Gilpin, M. 1994. Community-level competition: asymmetri-
cal dominance. Proc. Natl. Acad. Sci. U.S.A. 91: 3252-
3254.
Gosz, J.R. 1992. Ecological functions in a biome transition
zone: translating local responses to broad-scale dynamics.
In: Hansen, A. J. & di Castri, F. (eds.) Landscape bounda-
ries: consequences for biotic diversity and ecological
flows, pp. 55-75. Springer-Verlag, New York, NY, US.
Gosz, J.R. 1993. Ecotone hierarchies. Ecol. Appl. 3: 369-376.
Hansen, A.J., di Castri, F. & Naiman, R.J. 1988. Ecotones:
what and why? In: A new look at ecotones: Emerging
international projects on landscape boundaries. Biol. Int.
Spec. Issue 17: 9-46.
Hansen, A.J., Risser, P.G. & di Castri, F. 1992. Epilogue:
biodiversity and ecological flows across ecotones. In:
Hansen, A.J. & di Castri, F. (eds.) Landscape boundaries,
consequences for biotic diversity and ecological flows, pp.
423-438. Springer-Verlag, New York, NY, US.
Harper, K.A. 1995. Effect of expanding clones of Gaylussacia
baccata (black huckleberry) on species composition in
sandplain grassland on Nantucket Island, Massachusetts.
Bull. Torrey Bot. Club 122: 124-133.
Hill, M.O. & Gauch, H.G. 1980. Detrended correspondence
analysis, an improved ordination technique. Vegetatio 42:
47-58.
Jose, S., Gillespie, A.R., George, S.J. & Kumar, B.M. 1996.
Vegetation responses along edge-to-interior gradients in a
high altitude tropical forest in peninsular India. For. Ecol.
Manage. 87: 51-62.
Kupfer, J.A. & Cairns, D.M. 1996. The suitability of montane
ecotones as indicators of global climatic change. Prog.
Phys. Geogr. 20: 253-272.
Leathwick, J. R., Wilson, G., Rutledge, D., Wardle, P., Morgan,
F., Johnston, K., McLeod, M. & Kirkpatrick, R. 2002.
Land environments of New Zealand. David Bateman, Auck-
land, NZ.
Leopold, A. 1933. Game management. Charles Schribner’s
Sons, New York, NY, US.
Lloyd, K.M., McQueen, A.A.M., Lee, B.J., Wilson, R.C.B.,
Walker, S. & Wilson, J.B. 2000. Evidence on ecotone
concepts from switch, environmental and anthropogenic
ecotones. J. Veg. Sci. 11: 903-910.
Luczaj, L. & Sadowska, B. 1997. Edge effect in different
groups of organisms: vascular plant, bryophyte and fungi
species richness across a forest-grassland border. Folia
Geobot. Phytotax. 32: 343-353.
Neilson, R.P. 1991. Climatic constraints and issues of scale
controlling regional biomes. In: Holland, M.M., Risser,
P.G. & Naiman, R.J. (eds.) Ecotones: the role of land-
scape boundaries in the management and restoration of
changing environments, pp. 31-51. Chapman & Hall, Lon-
don, UK.
Neilson, R.P., King, G.A., DeVelice, R.L. & Lenihan, J.M.
1992. Regional and local vegetation patterns: the responses
of vegetation diversity to subcontinental air masses. In:
Hansen, A.J. & di Castri, F. (eds.) Landscape boundaries:
Consequences for biotic diversity and ecological flows,
pp. 129-149. Springer-Verlag, New York, NY, US.
Norton, D.A. 1992. Disruption of natural ecosystems by bio-
logical invasion. In: Willison, J.H.M., Bondrup-Nielsen,
S., Drysdale, C., Herman, T.B., Numro, N.W.P. & Pollock,
T.L. (eds.) Science and the management of protected
areas, pp. 309-319. Elsevier, Amsterdam, NL.
Odum, E.P. 1983. Basic ecology. Saunders College Publish-
ing, Philadelphia, PA, US.
Petts, G.E. 1990. The role of ecotones in aquatic landscape
management. In: Naiman, R.J. & Décamps, H. (eds.) The
ecology and management of aquatic-terrestrial ecotones,
pp. 227-261. UNESCO, Paris, FR.
Pound, R. & Clements, F.E 1900. The Phytogeography of
Nebraska. 2nd. ed. Botanical Seminar, Lincoln, NE, US.
Pulliam, H.R. 1988. Sources, sinks, and population regulation.
Am. Nat. 132: 652-661.
Putz, F.E. 1984. The natural history of lianas on Barro Colo-
rado Island, Panama. Ecology 65: 1713-1724.
Puyravaud, J.P., Pascal, J.P. & Dufour, C. 1994. Ecotone
structure as an indicator of changing forest-savanna bounda-
ries (Linganamakki Region, southern India). J. Biogeogr.
21: 581-593.
Risser, P.G. 1995. The status of the science examining ecotones
– A dynamic aspect of landscape is the area of steep
gradients between more homogeneous vegetation associa-
tions. Bioscience 45: 318-325.
Schonewald-Cox, C.M. & Bayless, J.W. 1986. The boundary
model: a geographical analysis of design and conservation
of nature reserves. Biol. Conserv. 38: 305-322.
Shmida, A. & Ellner, S. 1984. Coexistence of plant species
590 Walker, S. et al.
with similar niches. Vegetatio 58: 29-55.
Shmida, A. & Wilson, M.V. 1985. Biological determinants of
species diversity. J. Biogeogr. 12: 1-20.
Stace, C.A. 1997. New flora of the British Isles, 2nd. ed.
Cambridge University Press, Cambridge, UK.
Stevens, G.C. 1992. The elevational gradient in altitudinal
range: an extension of Rapoport’s latitudinal rule to alti-
tude. Am. Nat. 140: 893-911.
Stoutjesdijk, Ph. & Barkman, J.J. 1992. Microclimate, vegeta-
tion and fauna. Opulus Press, Uppsala, SE.
Tansley, A.G. 1939. The British Islands and their vegetation.
Cambridge University Press, Cambridge, UK.
Tansley, A.G. & Chipp, T.F. 1926. Aims and methods in the
study of vegetation. The British Empire Vegetation Com-
mittee, and The Crown Agents for the Colonies, London,
UK.
van der Maarel, E. 1976. On the establishment of plant com-
munity boundaries. Ber. Dtsch. Bot. Ges. 89: 415-443.
van der Maarel, E. 1990. Ecotones and ecoclines are different.
J. Veg. Sci. 1: 135-138.
van Leeuwen, C.G. 1966. A relation theoretical approach to
pattern and process in vegetation. Wentia 15: 25-46.
Whittaker, R.H. & Naveh, Z. 1979. Analysis of two phase
patterns. In: Patil, G.P. & Rosenzweig, M.L. (eds.) Con-
temporary quantitative ecology and related ecometrics,
pp. 157-165. International Cooperative Publishing House,
Burtonsville, MD, US.
Whittaker, R.H., Gilbert, L.E. & Connell, J.H. 1979. Analysis
of two-phase pattern in a mesquite grassland, Texas. J.
Ecol. 67: 935-952.
Wiens, J.A. 1992. What is landscape ecology, really? Land-
scape Ecol. 7: 149-150.
Wilson, J.B. 1990. Mechanisms of species coexistence: twelve
explanations for Hutchinson’s ‘Paradox of the Plankton’:
evidence from New Zealand plant communities. N. Z. J.
Ecol. 13: 17-42.
Wilson, J.B. & Agnew, A.D.Q. 1992. Positive-feedback
switches in plant communities. Adv. Ecol. Res. 23: 263-
336.
Wilson, J.B. & King, W.McG. 1996. Human-mediated veg-
etation switches as processes in landscape ecology. Land-
scape Ecol. 10: 191-196.
Wolf, J.H.D. 1993. Diversity patterns and biomass of epiphytic
bryophytes and lichens along an altitudinal gradient in the
northern Andes. Ann. Mo. Bot. Gard. 80: 928-960.
Received 20 June 2001;
Revision received 18 February 2003;
Accepted 27 February 2003.
Coordinating Editor: R.H. Økland.