Ecological Indicators 7 (2007) 1–21
                                                This article is also available online at:
                                                    www.elsevier.com/locate/ecolind


                            0
    The definition of a new plant diversity index ‘‘Hdune’’ for
     assessing human damage on coastal dunes—Derived
         from the Shannon index of entropy H0
                       R. Grunewald *, H. Schubert
             University of Rostock, Biosciences, Albert-Einstein-Strasse 3, 18051 Rostock, Germany
          Received 23 May 2005; received in revised form 26 September 2005; accepted 27 September 2005



Abstract

  The presented paper shows that the common usage of H0 is not an appropriate index of plant diversity for studying
disturbance in plant communities. H0 is strongly dependent on evenness, whereas species richness and species density are not
adequately incorporated. Therefore, the common usage of H0 is only a measure for relative diversity in respect to the maximum
possible diversity for each community. Furthermore, using the abundance of species relative to the total abundance of all species
will result in the loss of information on species density (plant cover) when H0 is calculated and comparisons between disturbed
                                                       0
and undisturbed communities become difficult. Instead it could be shown that the newly introduced index Hdune , which uses the
species abundance (as coverage percentage) relative to the constant sampling area for calculation, will more clearly detect
changes in species richness and composition. Sample size (plot size) is always constant and information on species density
                        0
(coverage) is also incorporated in the index. Hdune is used to compare different coastal dunes along the southern Baltic Sea coast
of Germany and Poland which differ in the degree of disturbance from recreational activities. The two main changes observed
were damages because of trampling and a change in species composition, because of increasing nutrient levels. Two methods to
detect different types and levels of anthropogenic disturbance and stress from recreational activities on coastal dunes are
presented.
# 2005 Elsevier Ltd. All rights reserved.

Keywords: Plant diversity; Sample size; Dunes; Disturbance; Tourism; Baltic Sea



1. Introduction                            Townsend et al., 2002; Patil and Taillie, 1982).
                                    Since it is regarded as an appropriate measure
 The Shannon Index of entropy H0 is often used            for diversity the present study tests its applicability
in ecological studies (Mouillot et al., 2005;             for detecting anthropogenic changes in plant
                                    diversity on coastal dunes along the southern coast
* Corresponding author. Tel.: +49 381 498 6086.
                                    of the Baltic Sea (Fig. 1). Following the discussion of
  E-mail address: ralf.grunewald@biologie.uni-rostock.de       H0 an adapted index Hdune is presented which
                                                  0

(R. Grunewald).                            incorporates the parameter species density (plant

1470-160X/$ – see front matter # 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ecolind.2005.09.003
2                  R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21




Fig. 1. Pommeranian Bight and study sites (BM = intensive beach management/cleaning) (1 = Wolin, undisturbed; 2 = between Trassenheide
and Karlshagen, undisturbed (no BM); 3 = Karlshagen, Camping (BM); 4 = Karlshagen, northern resort (BM); 5 = near between Karlshagen and
site 6 (no BM); 6 = Parking, north of Karlshagen (BM); 7 = north of site 6 (no BM, only few visitors)).

cover relative to the plot-size) into the diversity           and relative abundance of species. How many species
calculations (Fig. 2).                         are there in a given set (of individuals, biomass, plant
  This paper is not aimed at giving a thorough review         cover per area) and how abundant are they (or
of literature on diversity indices, since several authors,       biomass, plant cover). Some indices are aimed at
e.g., Magurran (1988), Peet (1974) or Pielou (1966),          either measuring just species richness (e.g. Margalef)
already give good introductions into measuring             other measure only evenness (how evenly are the
diversity; nevertheless, it is necessary to state the          individuals distributed among the different species). A
two major components of diversity: Species richness           different type of indices tries to incorporate both




          Fig. 2. The definition of H0 and Hdune (including other parameters calculated using H0 or Hdune ).
                           0                            0
                R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21             3

components of diversity into a single value, e.g.         light (inter species competition) is not dominant on
Hurlberts PIE-index, Simpson’s diversity index or the       many dunes, but salt spray, mechanical stress (from
widespread index H0 by Shannon (Shannon and            moving sand), nutrient availability and water
Weaver, 1949). Some are weighted towards the more         supply are key factors for plant life (Maun, 2004;
dominant species (Simpson) others give rare species        Kooijman and Besse, 2002; Packham and Willis,
more weight in the index (H0 , Mouillot and Lepretre,ˆ       1997; Ellenberg, 1996; Hesp, 1991). Only on older
1999). However, according to literature, most indices       dunes, where salt spray, nutrients and water are no
– including H0 – make use of the relative abundance of       longer exclusive limiting factors, competition for
each species in respect to the total number of           space as well as light will take place and will also
individuals (biomass, cover) sampled. Thus 2 com-         affect species richness (Tilman’s ‘‘equilibrium model
munities with 5 species each would not differ in          of plant resource limitation’’ in Hobohm, 2000;
diversity if the first community would have 50           Tilman et al., 2001).
individuals evenly distributed among the 5 species          The most obvious short time dynamics in beach and
and only 5 individuals for the second (possibly          dune systems are caused by heavy storm surges, but
disturbed) community (each species is consequently         constantly blowing winds cause long time changes and
represented by only one individual). Additionally, a        most of the sand accumulation; therefore, aeolian
single species ‘community’ does not differ from a         processes are most important for dune development
sample without any species. Nevertheless, these          complemented by effects from dune vegetation and
communities are very different.                  water (Hesp, 1991). On prograding coastlines along
  Therefore, it is necessary to take a closer look at H0     the southern Baltic Sea coast new dune habitat is
and how it was developed. Shannon (Shannon and           constantly being formed by the continuous sand
Weaver, 1949) described H0 as an information index or       supply. Plants break the wind speed causing sand
an index of entropy, which was initially used to          mobilized by the wind to settle and to build up small
analyze and measure information. How much infor-          primary dunes and start primary succession on the
mation is stored in a given set of data (amount of         upper beach (Grime, 2002; Hesp, 1991). Over time
information, p. 8) and how much could potentially be        more sand is being accumulated and new species,
stored (capacity, p. 8). Much is said about ‘‘informa-       typical for secondary dunes start to appear (e.g. Hesp,
tion content’’, ‘‘capacity’’, ‘‘probability’’ and         1991; Ellenberg, 1996). Different stages of primary
‘‘entropy’’. Nothing is said about ecological questions      succession will appear along a beach land gradient and
such as ‘‘species density’’, ‘‘interspecies competi-        plant and soil succession continues: CaCO3 is being
tion’’, ‘‘biomass’’ or even ‘‘diversity’’.             leached out of the dunes, pH decreases, humus starts to
  Translating Shannon’s approach on information          accumulate, and tertiary dunes develop. A typical
content and degree of entropy into the ecological         natural zonation along the southern Baltic Sea coast
meaning of diversity, the paper will try to show that the     from the upper beach towards the land will lead from
present usage of H0 does not take species density (i.e.      young successional stages of vegetation with low
species presence/absence or abundance in respect to        species numbers to more mature stages with higher
habitat) into account and therefore fails to detect major     species numbers and only on brown dunes—after inter
changes in dune habitats due to disturbance. The          species competition (competition for limited
authors believe that species density is a major          resources) becomes the dominant factor for species
parameter for studying community structure and           presence or absence—species numbers mostly decline
diversity. This is especially true for the study of        (Hesp, 1991). Such effects of species competition and
disturbance and extreme habitats like dunes.            diversity were also found by Goldberg and Estabrook
  The harsh dunal environment with strong sun           (1998), who studied the competitive exclusion of
radiation, poor sandy soils, minimal water supply, salt      species in communities characterized by high numbers
stress and mechanical stress from moving sand is the        of individuals per unit area.
dominant factor leading to the presence of highly           Fig. 3 shows one typical and one unchanged
adapted specialized species which can tolerate these        transect through coastal dunes indicating different
conditions (Hesp, 1991). Competition for space and/or       beach and dune zones along the southern Baltic Sea
4                   R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21




Fig. 3. Two transects through coastal dunes along the southern Baltic Sea coast. (A) Transect changed by recreational activities—now typical for
large parts of the dune coasts along the southern Baltic Sea Coast in Germany. (B) Unchanged natural transect.

coast according to the definition by v. Nordheim and            coast and the hypothesis was that ‘natural diversity’
Boedeker (1998) in the Red List of Biotopes and              can be used as an indicator. For this study disturbance
Biotope Complexes of the Baltic Sea the Belt Sea and            was defined as any human impact (related to
Kattegat. Besides coastal defense measures and               recreational activities) which will result in changes
building activities, two main anthropogenic impacts            in species diversity and/or community structure,
on southern Baltic Sea dunes are frequently discussed           which is a broader definition than the one by Grime
in literature (e.g. Piotrowska and Gros, 1998; Knapp,           (2002), but close or the same to the one used by Pickett
1996; Isermann, 1995; Jeschke, 1985).                   and White (1985). Since it is known, that low levels of
                                      (human) disturbance (intermediate disturbance
 Mechanical damages through trampling will                hypothesis, Connell, 1975) may result in a rise in
 directly damage plants and the vegetation cover             diversity, the term ‘natural diversity’ was used. Hill
 is reduced or even destroyed (sand remobilization)            et al. (2002) showed that increasing levels of
 which will lead to secondary succession once the             anthropogenic impact correlate with increasing num-
 disturbance has stopped (Grime, 2002).                  bers of alien species. Therefore, species which are
 Eutrophication as a result of faecal deposits from            either alien species to Germany or Poland or alien to
 tourists or gardening activities in the vicinity of           the unchanged (natural) dune habitat were used as
 holiday resorts will dampen the extreme habitat             additional indicators for disturbance (human influ-
 conditions on the dunes and untypical ruderal plants           ence, hemeroby, Kowarik, 1999).
 have a chance to establish themselves—changing               After testing the applicability of H0 for field data
 the species composition. Grime (2002) suggests              from Wolin Island and Usedom Island (Poland and
 that’’ fluctuations in resource availability (e.g.            Germany), different adaptations of this index were
 nutrients) are the key factor controlling plant             developed. As a result it could be shown that H0 is not
 community invasibility.’’                        an adequate measure for anthropogenic disturbance in
                                      coastal habitats, especially dunes. In the end an
  The aim of the study was to analyze the extent of            alternative way of calculating H0 was developed which
anthropogenic disturbance along different parts of the           is indicated by adding the suffix ‘‘dune’’. The
                R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21                   5

applicability of this newly introduced diversity          and mostly campers use the beach. Intensive beach
measure for other types of habitats must still be         cleaning operations are carried out at site 3 and at site
proven in forthcoming research.                  4, which is in Karlshagen. This site is used by many
                                  holiday makers and daytime tourists who use the
                                  nearby parking areas. Site 5 is just north of the resort
2. Material and methods                      and no intensive beach cleaning is carried out. This
                                  site is mostly visited by holiday makers from the resort
2.1. Study sites and sampling methods               and some visitors from parking areas, but numbers are
                                  lower than in the resort itself or at site 6, which is
2.1.1. Study area                         adjacent to a large parking lot. Many day tourists and
  The presented data was gathered from 2002 to          overnight campers (also illegal camping in the dunes)
2004 on the German part of the Island of Usedom/          use the beach and intensive beach cleaning operations
Uznam and on the Polish Island of Wolin (Fig. 1).         are carried out. About 1 km north of site 6 is the
The temperate climatic conditions are very similar         remote site 7. Only few people walk the distance from
(yearly mean approx. 7.5 8C; yearly precipitation:         the parking lot and intensive beach cleaning is not
                     ¨
575 mm, Reinhard, 1951, 1962; Muller, 2004) and          carried out. Sites 1 and 3 were only used for studying
coastal exposition differ only little between the sites      tertiary dunes and sites 5 and 7 were only used for
(Fig. 1). Westerly winds prevail during most of the        studying primary dunes.
year, but storm surges with peak wind speeds mostly          In addition to the sites mentioned above, several
come from northeasterly directions from November to        other dunes along the islands of Usedom and Wolin
February (Hurtig, 1957). The type of sand building up       were also studied and altogether data from 429 plots
the dunes is also very similar in mineral composition       (249 plots from tertiary and 180 from primary and
as well as grain size (Isermann, 2001) and all sites are      secondary dunes) are used for some statistical analysis
along potentially prograding coastlines with a similar       of the calculated diversity parameters.
zonation (Fig. 3). All study sites should therefore share
a similar potential for dune vegetation.              2.1.2. Sampling methods
  Usedom and Wolin are located in the Polish–            The method by Braun-Blanquet (1964) and Bark-
German trans-border-region and tourism is one of the        man et al. (1964) was used to document plant species
few thriving industries (Seidel, 2001). Poland joined       presence and coverage. The minimum area was
the European Union in 2004 and growing numbers of         determined according to Dierschke (1994) and a
visitors are expected in the future, whereas tourist        constant plot-size of 16 m2 was chosen for tertiary
numbers and tourist infrastructure are already very        dunes (grey dunes) and 4 m2 for primary and
high in Germany. Since large stretches along the          secondary dunes.1 The analysis of tertiary dunes
Wolin coast (Poland) are only accessible by foot or        was separated from the study of primary and
bicycle, these areas remain in a fairly undisturbed        secondary dunes. According to Braun-Blanquet
ecological state (Grunewald and Łabuz, 2004) and          (1964) only homogenous plots were studied so that
coastal recreation in northwestern Poland is concen-        any effects from different habitat conditions within
trated mostly around the holiday resort of Miedzyz- ˛       one plot could be eliminated (habitat diversity
             ´    ´
droje and the city of Swinoujscie. Fig. 1 shows the        influences species diversity). The method of doc-
seven different study sites which were chosen because       umenting each species and its coverage can be seen as
of expected and observed differences in recreational        a ‘‘non-destructive estimate of biomass’’ (Tilman
activities. Sites 1 (on the central part of Wolin between     et al., 2001) and is a common procedure to document
           ´    ´
Miedzyzdroje and Swinoujscie) and 2 (a remote beach
  ˛                               plant communities and study diversity patterns. Since
and dune area between the holiday resorts of            mosses and lichens are both on the same trophic level
Karlshagen and Trassenheide) both represent areas          1
                                    Isermann (1997) found similar values: even 4–8 m2 may be
which are only accessible by foot or bicycle (over         sufficient for some plant communities on tertiary dunes. For sec-
2 km from the nearest parking area or hotel). Site 3 is      ondary dunes it is important to have plots parallel to the coastline
adjacent to a large campground south of Karlshagen         (even 1 m2 may then be a sufficient sample size).
6                R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21

as vascular plants and make up a fair amount of plant
diversity on dunes, pedogenic lichens and mosses
were included in the study.
  As stated above, the analysis of primary and
secondary dunes was separated from the analysis of
tertiary dunes. The methods for plot selection also
differ between tertiary (older dunes) and secondary
and primary dunes (younger dunes), hence both
methods are described separately.

2.1.3. Primary and secondary dunes
  The vegetation of primary and secondary dunes
was sampled on Usedom Island (in Germany, Fig. 1) at        Fig. 4. View of the primary/secondary dunes and beach north of
the sites 2, 4, 5, 6 and 7. Between these sites visitor      Karlshagen (April 2003). At the base of the secondary dunes, the
numbers as well as the level of beach cleaning           small dune fence protecting the secondary dunes is visible. (1) The
operations and beach control (e.g. life guards) differ a      area immediately below the dune fence is the most popular spot for
                                  sunbathers. (2) Only in some distance from the dune fence damage
lot.                                from trampling is low and primary dune vegetation can develop
  Along the German Baltic Sea coast most secondary        (photo taken in April). (3) Intensive mechanical beach cleaning will
dunes are protected by a small fence at the base of the      remove debris washed ashore (including diasporas, rhizome). All
secondary dune and no trespassing is allowed; since        plants are destroyed and small primary dunes are leveled (right
these fences are always positioned in the same           foreground). (4) In the area of intensive beach cleaning, sand is blow
                                  across the bare beach and partly accumulates on the secondary
distance from the dune (between the secondary dune         dunes.
and the zone where primary dunes may form), it can be
used as a reference line (Figs. 3 and 4). These fences
are constructed from wooden poles (4 m apart, 80 cm        would not adequately reflect the mosaic vegetation
high) and a single wire so that any effect on sand         pattern on the beach.
accumulation initiated by the fence should be
minimal. Since it is always positioned at the base         2.1.4. Tertiary dunes
of the secondary dune (where sand accumulates to           The vegetation of tertiary dunes was sampled on
form new primary dunes) the fence was used to ensure        the German part of Usedom Island as well as on
that the same dune zones are compared with each          Wolin Island (Poland, Fig. 1). Site 1 on Wolin Island
other. Even though the fence is largely respected,         was chosen as an example for tertiary dunes with
people do not hesitate to cross the fence and defecate       almost unchanged typical dune vegetation (Grune-
on the dunes. The common effect observed was that         wald and Łabuz, 2004). The other sites for tertiary
along coasts with some recreational activities the         dunes were chosen, because the adjoining beaches
dunes are damaged in a mosaic pattern. Small pockets        represent different levels of recreational use, which
of bare sand (from trampling and sunbathing tourists)       should also lead to different levels of impact on the
alternate with areas covered with dense vegetation.        dunes. Nevertheless the impact on dunes was not
The objective was to study this pattern and use the        directly measured, but estimated using the para-
differences between sites as well as between the upper       meters accessibility, control of the beach (e.g. life
and lower plots as indicators for recreational activities.     guards), and the number of visitors on the beach (not
Therefore, at each site a 30 m long representative         on the dunes).
stretch of the fence, typical for this part of the beach,       Similar to the situation on primary and secondary
was marked with measuring tape and every 2 m plots         dunes, heavy mechanical damages to the vegetation
of 2 m  2 m (on each side of the measuring tape/         through trampling are clearly visible on tertiary dunes.
fence) were documented. The resulting 30 plots (15         They often alternate on a small scale with almost
behind, 15 in front of the fence) covered the variability     unaffected areas. Because of plant succession along
of the vegetation within this site. Any larger plot-size      the beach-land gradient (Isermann, 1997; Hesp, 1991)
               R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21               7

the obvious method would be to document plots from        2.2. Calculations
each stage of dunal plant succession and compare each
dunal zone from each study site with the same dunal         The following parameters were calculated from the
zone from the other sites. However, the dunes and each      gathered field data. Calculation of the diversity index H0
                                               0
dunal zone do not have exactly the same width. The        and the adapted index Hdune (Fig. 2) was done in order to
identification of each zone is difficult, because the       compare their sensitivity for changes on dunes. Since
plant community growing or the soil condition in this      evenness is often used as a parameter for community
area may already be the result of human disturbance:       structure (independent of species richness), the
                                             0
mechanical damages may be interpreted as a setback        calculation using Hdune is also presented. For its
in succession and the resulting secondary succession       calculation it is necessary to first calculate the average
may lead to vegetation which is still a typical dune       species coverage in each community and then the
                                              0
community—only growing further inland (high           maximum value for Hdune for this combination of total
                                                    0      0
naturalness, high hemeroby, Kowarik, 1999). The         cover and number of species (Hdune-max ). Hdune will
                                                0
human induced rise in nutrients may result in an         reach a maximum value (Hdune-max ), if all species are
accelerated succession, the resulting vegetation again      equally abundant. Similar to the calculation of E (using
may be a typical dune community—only growing           H0 ), Edune is the ratio between Hdune and Hdune-max .
                                                   0      0

closer to the beach than normal (high naturalness, high       Besides overall diversity, qualitative changes in
hemeroby). Therefore, it was decided not to compare       species composition were studied as well. It was
exactly the same dune zone from different sites with       expected that the number of ubiquist or ruderal species
each other.                           will increase as a result of rising nutrient levels on
  Instead the following protocol to choose the         tertiary dunes in areas with a high level of tourism.
individual plots was used: Three transects were         Neighbouring effects from nearby public parks and
documented at each study site. Each transect started       private gardens in the vicinity of holyday resorts were
on the youngest grey dunes with a 16 m2 plot and the       also expected to influence species diversity (species
last plot was made on the oldest non-forested dunes.                  ¨
                                 pool hypothesis, Partel et al., 1996) so that a rise in
One transect covered the areas exhibiting the highest      overall diversity may very well be a result of increasing
degree of damage (e.g. large illegal footpaths, fire       levels of human influence. The degree of natural
sites), one with intermediate damage, and one with        diversity (N) was calculated by classifying the present
the lowest level of damage. The study mainly           species into typical native dune species and untypical
focused on the open dunes, but single bushes or trees      dune species alien to the extreme dune habitat (including
                                                        0
were incorporated into the study. Very steep south-       truly alien species). The diversity index Hdune was then
ern slopes with naturally occurring erosion (strong       calculated twice: with all present species and without the
sun radiation in combination with high surface          species alien to the dunes (the later can be seen as a
temperatures and lack of water, Grunewald and          measure of natural diversity). The ratio between the two
Łabuz, 2004; Hesp, 1991) were also excluded from         can now be seen as the degree of natural diversity.
the investigation. The number of plots per study site      Species which are untypical for undisturbed dunes
may vary, because of the different widths of the         cannot cope with the extreme habitat conditions—once
dunes at each study site. However, three complete        they appear, they indicate change. Increasing nutrient
transects from young grey dunes to the older non-        levels are a key factor for the invasibility of habitats for
forested dunes were documented and grouped for          ruderal species (Grime, 2002). Similar to the calculation
each of the five sites. Since all sites should share the     of evenness, the ratio will be independent of species
same potential for phyto-diversity, this method         diversity and only a maximum value of one can be
ensured that the variability of the different dune        reached (if no species are excluded = complete natural
zones was documented and problems identifying          diversity). The following species were excluded from
certain dune zones did not occur. It also ensured that      calculating natural diversity: Acer platanoides, Acer
the variability of the different levels of human         pseudoplatanus, Aira praecox, Artemisia vulgaris,
disturbance through trampling was reflected in the        Campylopos introflexus, Conyza canadensis, Crataegus
different plots.                         spec., Elaeagnus angustifolia, Elaeagnus commutata,
8                  R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21

Prunus serotina, Rosa rugosa, Sambucus nigra, Sedum           secondary dunes are quite low and all species are
sempervivum, Solidago virgaurea, Symphoricarpos             named with their respective coverage (%). Even
albus, and Tortula ruraliformis. Most of the chosen           though large differences in total and individual
species are also alien species to Poland or Germany, but         coverage exist between the plots, the small differences
also some native species alien to the dunal habitat were         in evenness values result in almost similar H0 values.
excluded. The classification was based on habitat             Only one plot (4) shows a relative low value for H0 ,
requirements taken from literature (Ellenberg et al.,          which is caused by the dominance of Petasites spurius,
1992; Haeupler and Muer, 2000). The classification of           reflected in a low evenness. This community also has
the moss Tortula ruraliformis is based on personal            the highest total plant cover.
experience as well as a on a study by Berg et al. (1995).          The plots shown in Table 1 differ with respect to
  In order to compare the different diversity              anthropogenic impact. Whereas plots 2 and 4 (both at
parameters and the weight in calculating the diversity          site 2) showed no obvious signs of anthropogenic
indices (H0 , Hdune , E, Edune) correlation methods were
        0
                                     impact, visible signs of anthropogenic induced
applied. For the discrimination of dune sites, it was          mechanical stress (footprints, garbage) were observed
expected that damaged sites should significantly differ          at plots 1, 3 and 5 (all at site 4). Trampling obviously
from unchanged dunes. Therefore, data was also              resulted in reduced species numbers as well as
analysed using ANOVA like Kruskal–Wallis One-way             coverage, nevertheless, only small variations of H0
Analysis of Variance on Ranks (including all pair wise          and evenness could be found. Moreover, the undis-
multiple comparison procedures, p < 0.05).                turbed sites showed lower evenness values than the
                                     disturbed ones and caused a low value for H0 .
                                       Table 2 summarizes the most important diversity
3. Results and discussion                        parameters for three plots on one tertiary dune in
                                     Karlshagen (site 4). On undisturbed tertiary dunes
3.1. Application of H0 on dune vegetation                plant succession will lead to more mature plant
                                     communities than on primary and secondary dunes:
  Table 1 summarizes the results of five selected plots         more species are present and especially mosses and
from primary dunes. Species numbers on primary and            lichens play an important role within the ecosystem

Table 1
Examples of five different primary dune communities
Species                 Site 4 (Karlshagen) 09.09.2003                   Site 2 (Between Trassenheide
                                                      and Karlshagen) 07.09.2003
                    1           2           3           4            5
Signs of trampling           Yes          No          Yes          No           Yes
Ammophila arenaria           0.5          0.50         0.50
Festuca rubra              0.5          8.50                                1
Salsola kali              1.5          3.50         1.50                     0.2
Leymus arenarius            1.5          1.50         0.50          0.5          0.5
x Calammophila baltica         1.5          8.50         0.50
Lactuca tatarica            0.5          0.50         1.50
Petasites spurius                                             19.25
Summed coverage             6           23.00         4.50         39           2.2
Species number             6           6           5           3           4
H0                   1.661         1.367         1.465         0.753         1.250
 0
Hdune                  0.268         0.652         0.205         0.661         0.111
E                    0.927         0.763         0.910         0.685         0.902
 0
Hdune-max                0.276         0.750         0.212         0.796         0.114
Edune                  0.972         0.870         0.969         0.831         0.974
Numbers 1, 2, 3, 4 and 5 in the header represent plot numbers. All species and their coverage percentages (%) according to the scale from
Barkman et al. (1964) are shown.
                    R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21                   9

Table 2
Comparison of diversity indices on grey dunes with different levels of mechanical damages (level of impact)
Location                  Site 4 Karlshagen           Site 4 Karlshagen            Site 4 Karlshagen
Plot number                 6                   7                    8
Signs of trampling             Few                  Many                  Few
Total coverage               88.95                 25.5                  70.45
Number of species              24                  12                   23
 0
Hdune                    1.54526526              0.94030468               2.07258017
H0                      1.62013331              2.32097762               2.59164954
Edune                    0.52720795              0.95743649               0.84398109
E                      0.50978787              0.9340301                0.82655217
 0
Hdune-max                  2.93103557              0.98210659               2.45571874
For three neighbouring communities (site 12, holiday resort of Karlshagen) measured and calculated diversity parameters are shown. Areas with
a high level of trampling alternate with other areas not showing any signs of mechanical disturbance.


(Isermann, 1997; Hesp, 1991). The data in Table 2 was           mechanical stress was not measured directly, the
taken from site 4 (Karlshagen), because it reflects the          presence of illegal fire sites, garbage, and different
mosaic pattern of different levels of mechanical             footpaths supports this assumption. Since all plots are
influence on that particular dune. Even though               within the same part of the dune, they should have a
                                     similar potential for natural diversity and any
                                     differences are likely to be the result of anthropogenic
                                     change. Thus the plots shown are used as typical
                                     examples for the two extreme situations ‘‘heavy’’ and
                                     ‘‘little’’ mechanical disturbance.
                                       The effects of evenness and species richness on H0
                                     illustrate how it fails to detect different levels of
                                     mechanical disturbance. Similar to the results shown
                                     in Table 1, H0 mostly responds to changes in evenness.
                                     Changes in species richness are not reflected
                                     adequately, even though authors have favoured H0
                                     for its sensitivity in respect to rare species (e.g.
                                     Magurran, 1988). Plot number 7 (with 12 species and a
                                     very low coverage, because of trampling) has a much
                                     higher diversity according to H0 compared to plot
                                     number 6. The weight H0 puts on evenness becomes
                                     even more clear looking at plot number 8 which has
                                     roughly the same species richness as plot number 6,
                                     but a very high evenness value. Changes in coverage
                                     and species richness, which are both influenced by
                                     trampling, are not reflected by H0 (Fig. 5).

                                     3.2. Calculating and using H0 —community
                                     models

                                       To evaluate the weight H0 puts on evenness,
                                     examples of communities with different total cover-
Fig. 5. View of tertiary dunes. (a) Site 4 within the holiday resort of
Karlshagen (July 2003) showing heavy signs of trampling. (b) Site 2
                                     age, numbers of species and abundances were
between Trassenheide and Karlshagen (August 2003) with very few      constructed. Fig. 6 summarizes the main parameters
visitors.                                 of these theoretical communities and shows their
                                                                                       10
                                                                                       R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21




                               0
Fig. 6. Different community models for comparison of H0 and Hdune and how they are influenced by evenness, species richness and plant cover (Dmg = species richness index by
Margalef; Dsim = diversity index by Simpson).
                R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21                   11

species abundance patterns. Communities C1, C4 and         The specific protocol given by Haeupler (1982) for
C7 show how H0 decreases, if species richness           plant communities also calls for this procedure and
decreases—in these examples evenness is kept            information on species density or species coverage is
constant at its maximum (1). Looking at the            lost in the calculation of H0 .
community groups C1, C2, C3 or C4, C5, C6 or                     0
C7, C8, C9 the way H0 reacts to changes in evenness        3.3. The adapted Hdune —an attempt to overcome
(one or more species become dominant and evenness         these problems by theory
decreases) is analyzed. C2, C5 as well as C8 all show
decreasing H0 values respectively to C1, C4 and C7           Since the calculation of pi relative to the summed
even though there is no absolute loss in species. The       total coverage is responsible for the way H0 reacts to
total coverage actually increased. In the communities       changes in coverage, the index was again calculated
C3, C6 and C9 dominant species were deleted,            using the coverage percentage Pi-dune. This value is an
evenness again increased and so did H0 . Species          abundance or coverage estimate for each species
density (coverage) as well as species numbers           relative to the constant sample or plot size (and not
decreased, but H0 still ‘‘measures’’ a higher diversity.      to the summed total coverage of all species, Fig. 2). This
                                         0
  As stated already above, H0 very sensitively reacts       new index (Hdune ) should now follow the rationale that
on changes in evenness, whereas changes in total          when biomass and/or species richness decreases
coverage are not reflected. Hobohm and Petersen           substantially as a result of trampling, the index must
(1999) as well as Haeupler (1982) already stated the        not increase as a result of an increase in evenness, but
strong weight H0 puts on evenness. The most drastic        must decrease. In other words: A species with 5% cover
example for this is given by comparing C8 with C9.         in a plot will be rated the same in any community,
From C8 to C9 three species became ‘‘extinct’’           regardless of other species presence or absence.
without any other changes; however, H0 was higher in          Fig. 7a shows the effects of increasing coverage
C9 than in C8 where these species were the most          values and increasing evenness values for 100
abundant. Such species-poor communities are typical        theoretical communities. Species richness (10 species)
for many dunes (Table 1; Grunewald, 2004a,b).           is kept constant. Very unevenly distributed commu-
  This problem can be overcome in part by application       nities slowly increase in evenness along the z-axis
of an index focusing on species richness alone, e.g.        towards the back of the graph. From the left towards
Margalef (Dmg), which is also shown in Fig. 6.           the right side of the graph the plant cover increases
However, because coverage is not included at all,         along the x-axis and so does in most cases the index.
effects of mechanical disturbance as shown in Fig. 6        This is not true for very unevenly distributed
(e.g. from C2 to C1) would not be reflected. Simpson’s       communities (front row) which is due the fact that
            P                      the limit of LN(Pi-dune) is ‘‘0’’ and as soon as LN(Pi-
index (Dsim ¼ 1 À P2 ) is often used as an alternative
              i
to H0 , but it also cannot overcome this problem (Fig. 6).     dune) reaches values above ‘‘À1’’ (above a coverage of

By squaring the relative abundance of each species,        approx. 40% for the dominant species) the index will
dominant species also dominate the index (Dsim can be       slightly decrease again.2 Using a different base/
seen as a dominance index, Magurran, 1988); hence,         logarithm may partly solve this problem, but a
Simpson’s index also has a restricted sensitivity for       decrease will still occur. On the other hand the
                                                       0
species richness and the comparison of C7, C8 and C9        steepness of the curve (the weight Hdune puts on
show similar results as for H0 .                  coverage and/or evenness) will be different. This
  For the study of human change in species poor dunal       problem may be overcome by substituting the term
systems a fine resolution of both parameters, relative       LN(Pi) with [(LN(Pi)) À 1] which has a limit of ‘‘À1’’
abundance of each species and species number in any        (Fig. 7b). Since testing field data did not reveal
diversity index is required. In addition, a formula must      significant differences between both versions of the
be constructed which prevents the loss of information        2
                                    The logarithmic term will lead to similar effects in the calcula-
on total coverage (species density). The common way to       tion of H0 , since the second normalization may lead to equivalent
calculate H0 (Magurran, 1988) involves a normalization       values for pi as for pi-dune; however, the changes in evenness will
step by converting total abundances into relative (%).       cover up such an effect.
12                  R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21




                                        0                  0    0
Fig. 7. How evenness and increasing plant cover affect: (A) the adapted index Hdune and (B) the extreme version of Hdune ; Hdune and an extreme
       0
version of Hdune were analyzed using changing species abundance patterns. One hundred theoretical communities with a constant species
richness of 10 species were used. Evenness is being increased along the x-axis and coverage increases along the y-axis. The z-axis shows the
calculated diversity index and how it reacts to changes in coverage and evenness.


                                                           0
index (even in the species poor dune communities),             still an important part of the index, but Hdune is more
the simpler version was used in this study.                weighed towards species richness and coverage—
  Figs. 6 and 7 show that Hdune , in contrast to H0 ,
                 0
                                      evenness becomes important for discriminating bet-
reflects the effects of differences in coverage (species          ween communities roughly sharing the same number
                          0
density). Evenness (Edune) calculated using Hdune             of species and a similar total coverage. This is shown
remains independent from both species richness               by comparing the community pairs above and then
(n = 429 plots: r2 = 0.034)3 and coverage (r2 = 0.170).4          looking at the independent indices for evenness
      0
  Using Hdune , C15 now is rated highest since it has           (Edune) and species richness (Margalef). A single
the highest species richness and biomass/cover, the            species community will have a maximum Edune (1).
later can also be seen as an important element of
diversity as well as ecosystem stability (especially on          3.4. Application of the new formula on the field
dunes). Looking at values for evenness, both calcula-           data
tions (using H0 or Hdune ) give similar values and Edune
           0
                                                    0
does fulfill the two most important properties of the             Using the new index Hdune the data presented was
conventional E:                              again analyzed (Tables 1 and 2). It is now possible to
                                      clearly distinguish between the different sites and their
                                                             0
 A maximum value of one can only be reached if all            different levels of anthropogenic impacts. Hdune seems
 species have the same abundance.                     to clearly detect changes in species richness as well as
                                                          0
 Edune is independent of biomass/plant cover and             density (coverage). In the following Hdune was tested
 species richness                             and used to compare different sites from Usedom and
                                      Wolin Island.
  Very slight changes in evenness are detected by Edune          3.4.1. Results primary and secondary dunes
as well as by E (C13–15). Looking at C1, C10 and C13,             Figs. 8 and 9 show box and whisker plots for Hdune0
C11 and C14 and at C12 and C15 shows that evenness is              0
                                      and H from primary and secondary dunes on Usedom
 3                                    Island. The boxes represent the second and third
  Results are taken from plots from several dune sites on Usedom
and Wolin Islands, not only the sites presented in this paper.
                                      quartiles, the bar within the boxes indicates the median
 4
  Testing correlation of E (using H0 ) with species richness:      and whiskers indicate the smallest values within 1.5
r2 = 0.138 testing E (using H0 ) with coverage: r2 = 0.0128.        interquartile ranges of the bottom. Small dots indicate
                   R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21                     13




                                                       0
Fig. 8. Comparison of primary and secondary dunes with different levels of recreational activities using Hdune . The protective dune fence was
used as a reference line between the upper and lower plots. The boxes represent the second and third quartiles, the bar within the boxes indicates
the median and whiskers show the standard deviation. Small dots indicate extreme values (2 = between Trassenheide and Karlshagen,
undisturbed (no BM); 4 = Karlshagen, northern resort (BM); 5 = between Karlshagen and Parking (no BM); 6 = Parking, north of Karlshagen
(BM); 7 = north of Parking (no BM)).




Fig. 9. Comparison of primary and secondary dunes with different levels of recreational activities using the diversity index H0 (Shannon, 1949).
The protective dune fence was used as a reference line between the upper (A) and lower plots (B). The boxes represent the second and third
quartiles, the bar within the boxes indicates the median and whiskers show the standard deviation. Small dots indicate extreme values
(2 = between Trassenheide and Karlshagen, undisturbed (no BM); 4 = Karlshagen, northern resort (BM); 5 = between Karlshagen and Parking
(no BM); 6 = Parking, north of Karlshagen (BM); 7 = north of Parking (no BM)).




extreme values which fall outside the whiskers. The              The comparison between the sites is the second
mean (m) and the number of plots documented at each             method of discriminating the sites and assessing
site (n) are also given in the figures.                   human impact: Table 3A and B shows the results of a
  A statistical analysis (t-test, Kruskal–Wallis One-           one-way ANOVA on ranks for both H0 and Hdune .
                                                            0

way Analysis of Variance on Ranks, all pairwise
multiple comparison test) shows that there are               3.4.2. Results tertiary dunes
significant differences ( p < 0.05) between and within              Figs. 10 and 11 show the use of Hdune and H0
                                                            0

the sites (above and below the dune fence). Both H0 as           comparing a total of 119 plots from 5 different sites on
      0
well as Hdune detect significant differences between             Usedom and Wolin Islands (Germany and Poland).
                        0
the upper and lower plots in most sites, only Hdune does          The box and whisker plots reveal the main character-
so at the site ‘‘Northern Beach’’.                     istics of the different sites including mean (m) and
14                  R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21

Table 3
Results of a one-way ANOVA on ranks for both H0 (A) and Hdune (B) for primary and secondary dunes
                             0




Significant differences are indicated with ‘‘yes’’. Insignificant differences are indicated with ‘‘no’’ ( p < 0.05).




number of analysed plots per site (n). The boxes              fall outside the whiskers. The mean (m) and the
represent the second and third quartiles, the bar within          number of plots documented at each site (n) are also
the boxes indicates the median and whiskers indicate            given in the figures.
the smallest values within 1.5 interquartile ranges of             Mean values show strong differences between the
the bottom. Small dots indicate extreme values which            sites, and a statistical analysis (t-test, Kruskal–Wallis
                    R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21                       15




Fig. 10. Comparison of tertiary dunes with different levels of         Fig. 11. Comparison of tertiary dunes with different levels of
                0
recreational activities using Hdune (n = number of plots per site,       recreational activities using H0 (n = number of plots per site,
m = mean value per site) (1 = Wolin, undisturbed; 2 = between         m = mean value per site) (1 = Wolin, undisturbed; 2 = between
Trassenheide and Karlshagen, undisturbed (no BM); 3 = Karlsha-         Trassenheide and Karlshagen, undisturbed (no BM); 3 = Karlsha-
gen, Camping (BM); 4 = Karlshagen, northern resort (BM);            gen, Camping (BM); 4 = Karlshagen, northern resort (BM);
6 = Parking, north of Karlshagen (BM)).                    6 = Parking, north of Karlshagen (BM)).


One-way Analysis of Variance on Ranks, all pairwise              Fig. 2). The maximum value for N is one, since
multiple comparison test) reveal significant differ-              the natural diversity cannot be higher than the
ences shown in Table 4.                            complete diversity using Hdune (using H0 there may
                                                     0

  The second important aspect for dune conserva-               be an increase in natural diversity, if untypical
tion (tertiary dunes) is species composition: How               species are left out of the calculation like it was
many untypical species occur on the dunes and how               done in Fig. 6: C2 and C3, C5 and C6, C8 and C9).
dominant are these species? This degree of natural               Significant differences between sites are shown in
diversity (N) is shown in Fig. 12 (formula given in              Table 4.
Table 4
Results of a one-way ANOVA on ranks for tertiary dunes




In the upper right half results from both H0 (A) and Hdune (B) are shown. In the lower left half results for the degree of natural diversity are given.
                           0

Significant differences are indicated with ‘‘yes’’. Insignificant differences are indicated with ‘‘no’’ ( p < 0.05).
16                    R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21

                                        r2 = 0.16, Pearson correlation coefficient = 0.788)
                                        shows that species richness does not correlate as
                                        strongly with H0 on tertiary dunes (high number of
                                        species, high evenness, high coverage) than it does on
                                        primary and secondary dunes (low number of species,
                                        low coverage). Evenness seems to be the dominating
                                        factor in communities with low coverage values which
                                        was also shown in Tables 1 and 2.
                                          Total coverage exhibits a strong correlation with
                                        species richness (n = 429, r2 = 0.71, Pearson correla-
                                        tion coefficient = 0.881) and cross-correlates with H0
                                        as well (n = 429, r2 = 0.55, Pearson correlation
                                        coefficient = 0.747). This may be explained by the
                                        fact that higher coverage values are mostly found on
Fig. 12. Comparison of tertiary dunes with different levels of recrea-     tertiary dunes which also have higher species richness
tional activities using the degree of typical dune species (natural      than primary or secondary dunes with lower coverage.
diversity) present in relation to the overall diversity (Fig. 9). This ratio
                                        The correlation between cover and H0 is much weaker,
was used as a measure of the degree of natural diversity (N) (n = num-
ber of plots per site, m = mean value per site) (1 = Wolin, undisturbed;    when only data from primary and secondary dunes is
2 = between Trassenheide and Karlshagen, undisturbed (no BM);         used (n = 180; r2 = 0.07, Pearson correlation coeffi-
3 = Karlshagen, Camping (BM); 4 = Karlshagen, northern resort         cient = 0.327) or only data from tertiary dunes
(BM); 6 = Parking, north of Karlshagen (BM)).                 (n = 249, r2 = 0.09, Pearson correlation coeffi-
                                        cient = 0.330). This shows that information on species
3.4.3. Discussion                               density is being lost in the calculation of H0 .
Main properties of H0 and Hdune0
                                          The correlation between H0 and Hdune was also
                                                             0
       0    0
  Testing H and Hdune and the parameters important              tested using data from primary, secondary and tertiary
for biodiversity (species richness, evenness, and               dunes and revealed a strong correlation of the two
species density) were conducted using 429 different              tested parameters (n = 429, r2 = 0.77, Pearson corre-
plots from primary, secondary and tertiary dunes on              lation coefficient = 0.876). However, when only
Usedom and Wolin Island (including data from sites               tertiary dunes or only secondary and primary dunes
not presented in this paper). Correlation coefficients             are being compared, the correlation differs a lot
between these three parameters and Hdune as well as H0
                   0
                                        (tertiary dunes: n = 249, r2 = 0.50, Pearson correlation
were calculated.                                coefficient = 0.705 and primary and secondary dunes:
  As expected using all 429 plots evenness and                n = 180; r2 = 0.25, Pearson correlation coeffi-
species richness both correlate with H0 (evenness:               cient = 0.498). In the species poor communities with
r2 = 0.21, Pearson correlation coefficient = 0.714 and             low coverage the differences between the two indices
species richness: r2 = 0.77, Pearson correlation coeffi-            become evident. Once plant succession has led to
cient = 0.920). However; the correlation shows                 more complex plant communities with higher cover-
important differences when data from tertiary dunes              age values both indices start to correlate more. It is
and secondary and primary dunes are analysed                  obvious, that the closer the summed total coverage is
separately: For species poor secondary and primary               to 100%, the more similar H0 and Hdune will be. This
                                                           0

dunes (n = 180, r2 = 0.75, Pearson correlation coeffi-                              0
                                        shows that the present use of H may be restricted to
cient = 0.882) the correlation of evenness with H0 is             saturated communities (high coverage) where inter
much stronger than for species rich tertiary dunes               species competition is the dominant factor for species
(n = 249, r2 = 0.136, Pearson correlation coeffi-                presence or absence. Whenever total cover is not close
cient = 0.766). The separate analysis of correlation              to 100% H0 will be dominated by the factor evenness.
between species richness and H0 (secondary and                   Peet (1974) calls H0 a heterogeneity index and
primary dunes: n = 180, r2 = 0.76, Pearson correlation             Haeupler (1982) and Hobohm (2000) criticise the
coefficient = 0.877 and for tertiary dunes: n = 249,              strong weight of evenness in calculating H0 and do not
                R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21             17

propose its broad use as a general diversity indicator.      a straightforward approach for comparing diversity
Instead they propose to use evenness, because it is        between different sites and not even the same site at
independent of species richness and calculate typical       different times, because the sample size is not kept
evenness values for different communities. The strong       constant.
influence of evenness on H0 was shown by their strong             0
                                   For Hdune a different way was chosen to make sure
correlation. Nevertheless, the problem of species         the number of available choices (=sample size) will be
density has been largely neglected and only little         kept constant. Magurran (1988) describes common
information was found in literature (e.g. Pielou, 1966).      definitions of sample size.
Instead, in many studies H0 and absolute cover are
calculated and analysed separately (e.g. Rodgers,          Number of individuals sampled (or coverage/
2002; Andersen, 1995).                        biomass).
  Because of the strong weight of evenness, H0 is only       Area of sampling.
a measure for relative diversity in respect to the          Time for sampling.
maximum possible diversity for one specific commu-
nity at one specific time (see also Shannon and           Studying plant communities the habitat (minimum
Weaver, 1949; p. 8: ‘‘To be sure, this word information      area) can be completely sampled so that time as a
in communication theory relates not so much to what        measure of sample size (sampling effort) can
you say, as to what you could say. That is, information      theoretically be neglected.
is a measure of one’s freedom of choice when one            The remaining two parameters (coverage and area)
selects a message.’’). According to Shannon ‘‘, the        can each be seen as one dimension of sample size
amount of information is defined, in the simplest          using H0 . It is an important difference, if one samples
cases, to be measured by the logarithm of the number        200 individuals in an area of 100 m2 or 200 individuals
of available choices . . .’’ (Shannon and Weaver, 1949;      are found within 10 m2. It is also a huge difference, if
p. 9). Hence, there is the need to define the number of       100 individuals or 200 individuals are found within
available choices (=sample size).                 10 m2. In other words: Information on species density
  So far, using H0 in biological diversity studies this      and on species presence or absence is important.
                                               0
was the total number of individuals sampled or the           The different way Hdune is calculated does not omit
summed total coverage (amount of biomass) of all          the parameter species density, but defines it as a
plant species, Haeupler, 1982; Magurran, 1988;           parameter of species diversity itself. For each species,
Townsend et al., 2002). Whenever total coverage is         its density is not related to the other species present,
used as the definition of sample size, then this number       but to the habitat it grows in. Thus, the sample area can
of available choices must be kept constant. Commu-         be defined as the sole basis of comparison (=‘‘number
nities saturated with species and individuals (inter-       of available choices’’, Shannon and Weaver, 1949).
species competition as the driving force of species        The inclusion of coverage (species density) into the
                                            0
presence or absence) may roughly meet this require-        calculation of Hdune creates a counterweight to the
                                                      0
ment since the total cover will be around 100%. If         very dominant factor evenness. In Hdune , evenness is
disturbance like trampling leads to a reduction of total      used to differentiate between communities sharing
cover, the sample size has substantially changed          similar numbers of species and comparable coverage
making an analysis of disturbance using H0 very          values (Table 2). Evenness (Edune) can be easily
difficult. As shown in Tables 1 and 2, the total plant       calculated in a similar way as E (evenness using H0 ,
cover or species density is not constant, but varies a lot     Fig. 3). It is independent from both total coverage and
as a result of human impact. In Table 2 plots 6 and 8       species richness and is a much better index for
roughly share the same species number, but differ in        analyzing community structure than H0 as it was
evenness and cover, because of damage from             already shown by Haeupler (1981). Table 2 as well as
trampling in plot 8. The information on species          Fig. 6 show that E and Edune differ very little. The main
density is being lost once the abundance of species        differences are single species stands: H0 does not
relative to summed total coverage is calculated and        differentiate between no species or single species
evenness will dominate the index. Therefore, H0 is not       stands and cannot be calculated in such situations.
18                R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21

 0
Hdune can be calculated for single species stands and        composition were observed especially in tertiary
Edune will have a value of one. Once these single          dunes. Similar findings were made by Rodgers
species stands are left out, E (evenness using H0 ) and       (2002) and Piotrowska (1988). Specialized species
         0
Edune (using Hdune ) strongly correlate (n = 397,          are driven out because of a rise in nutrients and an
r2 = 0.77, Pearson correlation coefficient = 0.880).         invasion of ubiquist species. Therefore a classification
             0
This enables users of Hdune to also use Edune for their       of the species with respect to natural diversity and
community analysis.                         naturalness (N) was needed. Such a classification of
                                  species is a common procedure in ecological and
     0                                                         ´
Using Hdune for the detection of human changes on          conservational studies (e.g. Grime, 2002; Garcıa-
coastal dunes                            Mora et al., 2000, 2001; Martinez and Psuty, 2004;
         0
  Applying Hdune on the field data, the greater           Rodgers and Parker, 2003; Rodgers, 2002). The rise in
         0
sensibility of Hdune for changes is revealed for primary      more ubiquist or ruderal species may be explained by a
and secondary dunes at the most northern site (7),         rise in nutrients on tertiary dunes and by gardening
where only few heavily trampled pockets of bare sand        activities in the vicinity of holiday resorts like
                   0
exist, which are only detected by Hdune. One important       Karlshagen (site 4). Neighbouring effects from public
            0    0
difference between H and Hdune is the value for mono-        and private gardens adjacent to dunes may be another
species stands: In such cases Hdune , in contrast to H0 , is
                  0
                                  explanation. Habitats can only be settled by species
able to differentiate with respect to their different        which are able to reach this habitat, as was shown by
coverage.                               ¨
                                  Partel et al. (1996). The species pool of the area around
  The statistical analysis between different secondary      the studied habitat (regional species pool) influences
dunes shows that H0 does not detect any significant         species diversity in the habitat itself. According to the
differences between sites behind the dune fence (Fig. 9       species pool hypothesis, holiday resorts would have a
             0
and Table 3). Using Hdune the low coverage values are        larger species pool and influence the diversity on the
not neglected and changes between the remote sites         adjacent dunes; nevertheless both explanations are
(less people = less disturbance) 2 and 7 on one side        linked to tourism or anthropogenic change in general
and the heavily used sites in and around Karlshagen         and both are likely to jointly influence species
are significant (Fig. 8 and Table 3). In front of the dune      diversity on the dunes. The low number of ruderal
fence the situation becomes clearer and the effects of       species at site 3 (campground south of Karlshagen)
tourists and beach management are obvious: H0            and site 6 (Parking) may indicate the importance of the
detects the great differences between the undisturbed        species pool, but the level of control at site 3
                      0
sites and the heavily used sites, but Hdune detects a        (campground with life guards on the beach and
further difference between site 2 and the almost bare        campground staff) is quite high, so that not many
sand at site 4 (Karlshagen, Resort Center).             people trespass the dunes. Public toilets are available
               0
  The mean values of Hdune can also be used to further      at site 3 as well, so that the level of disturbance on the
discriminate the sites. Above and below the dune          dunes is generally quite low. At site 6 the situation is
                 0
fence in site 4 (Karlshagen) Hdune detects damage from       similar: the dunes are easily overlooked from the
trampling which results in a low mean. On the other         access pathways to the beach and only very few bushes
hand the mean values in site 2 (between Trassenheide        and trees provide some shelter for tourists defecating
and Karlshagen) are slightly higher than at ‘‘Northern       on the dunes. Most will walk further into the coastal
Beach’’.                              pine forest, where a number of ruderal species indicate
              0
  For tertiary dunes Hdune detects significant changes       higher nutrient levels and eutrophication (e.g. Sam-
between site 1 on Wolin (undisturbed) and site 2          bucus nigra, Urtica dioica). These areas in the coastal
(between Trassenheide and Karlshagen) on one side          forest however were not studied.
and the heavily disturbed sites in and north of the           For tertiary dunes, mosses and lichens have been
holiday resort of Karlshagen. H0 was only able to          shown to be major components of plant diversity,
differentiate for 50% of those sites.                Piotrowska (1979) states that 50% of the total number of
  Aside coverage effects, mostly caused by mechan-        plant species are either mosses or lichens. In the present
ical damage, qualitative changes in plant community         study they accounted on average (!) for 44% of the
                R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21                  19

diversity on grey dunes (including disturbed plots).        detection of anthropogenic changes on dunes showed
                                                    0
Haeupler (1981) does not include mosses and lichens        the applicability of this index Hdune and therefore
into his analysis and uses the following two arguments.      verifies the assumption that coverage (species density)
                                  should be included into diversity studies. The new
1. The level and quality of species identifications         index has been shown to allow for calculation of
  varies between scientists and more problematic         evenness (Edune) in a similar way as H0 .
  groups should therefore be excluded.                A second potential degradation factor for dunal
2. Even though cryptogams make up an important           ecosystems is a human induced rise in nutrients and or
  amount of diversity they often create a separate        neighboring effects from nearby gardens and parks,
  vegetation layer and should be analysed separately.       leading to increased plant coverage and invasion of
                                  ubiquist or ruderal species. Such effects have been
Since in this study species identification was                                 ¨
                                  described in literature (Cornell, 1999; Partel et al.,
conducted by one single scientist and critical species       1996), but cannot be detected by means of H0 or Hdune
                                                            0

were crosschecked by experts the first argument can be       alone and therefore require the use of a classification
neglected for this study. An alternative to circumvent       system, discriminating between ‘‘typical’’ and ‘‘non
the problem would be pooling of all pedogenic lichens       typical’’ dune species. The resulting calculation of the
and separately all mosses into two functional groups,       degree of natural diversity in the different dune
giving not the same, but clearly better results than        communities (naturalness = N) revealed qualitative
neglecting these organisms. According to Piotrowska        changes in dunal species composition.
                                           0
(1979) cryptogamic plant communities show little           The use of Hdune may not be restricted to coastal
resistances against mechanical stress and moreover         habitats and seems likely to be applicable in other
have only little power of fast regeneration; hence, it is     habitats as well. So far, this has not been tested.
very important to include them in the study. The
second argument can also be countered by proposing
to calculate indices according to ‘‘functional groups’’      Acknowledgements
(grasses, herbs, dwarf shrubs, other shrubs, C4
species, etc.) as done successfully for dunes, other         The authors would like to thank Tomasz A. Łabbuz
terrestrial communities or aquatic habitats by, e.g.        from the Institute of Marine Sciences, University of
   ´
Garcıa Novo et al., 2004; Hellemaa, 1998; Garcıa-  ´       Szczecin for cooperation and help during the field-
Mora et al., 2000, 2001.                      work in Poland. The authors would also like to thank
                                  federal and local state institutions for granting access
                                  to dunes and the German Federal Environmental
4. Conclusions                           Foundation (DBU) for financial support.

  The present study shows that the common
assumption that H0 (species number and abundance          References
relative to the total abundance) is an appropriate
estimate for diversity cannot be followed for           Andersen, U.V., 1995. Resistance of Danish coastal vegetation types
unsaturated and/or disturbed systems with low plant          to human trampling. Biol. Conserv. 71, 223–230.
cover and/or changing cover. In these calculations the       Barkman, J.J., Doing, H., Segal, S., 1964. Kritische Bemerkungen
                                          ¨
                                   und Vorschlage zur Quantitativen Vegetationsanalyse. Acta Bot.
sample size is not kept constant. H0 is a measure of         Neerlandica 13, 394–419.
diversity relative to the maximum diversity possible        Braun-Blanquet, J., 1964. Pflanzensoziologie. Grundzuge der ¨
for the number of species present in each site and          Vegetationskunde, 3rd ed. Springer, Berlin, Wien, New York,
hence in strongly dominated by evenness. Therefore,          pp. 865.
H0 is not sensitive for changes, because of human         Berg, C., Schramm, J., Dieminger, J., 1995. Die Moosflora des
                                   Nationalparks Vorpommersche Boddenlandschaft in ihrer
influence. Incorporating coverage as a term influen-                 ¨
                                   Bedeutung fur Pflege und Entwicklungskonzepte—Zusammen-
cing diversity, as done in the present case, overcomes        stellung erster Ergebnisse. Study for the National Park Vorpom-
this problem. Applying this new concept on the            mersche Boddenlandschaften. Unpublished.
20                   R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21

Connell, J.H., 1975. Some mechanisms producing structure in        Hill, M.O., Roy, D.B., Thompson, K., 2002. Hemeroby, urbanity and
  natural communities: a model and evidence from field experi-        ruderality: bioindicators of disturbance and human impact. J.
  ments. In: Cody, M.L., Diamond, J. (Eds.), Ecology and Evolu-       Appl. Ecol. 39, 708–720.
  tion of Communities. Harvard University Press, Cambridge,                       ¨
                                     Hobohm, C., 2000. Biodiversitat. Quelle und Meyer. UTB, Wiebel-
  Mass, pp. 460–490.                             sheim, pp. 8–21.
Cornell, H.V., 1999. Unsaturation and regional influences on species    Hobohm, C., Petersen, J., 1999. Zur Artenvielfalt von Zwergbin-
  richness in ecological communities: a review of the evidence.       sengesellschaften. Mitt. Bad. Landesver, Naturkunde und Nat-
  Ecoscience 6 (3), 303–315.                         urschutz, N.F. 17, 313–318.
Dierschke, H., 1994. Pflanzensoziologie: Grundlagen und Metho-       Hurtig, T., 1957. Physische Geographie von Mecklenburg. In:
  den. Ulmer Verlag, Stuttgart, pp. 683.                   Greifswald, VEB, Deutscher Verlag der Wissenschaften, Berlin,
Ellenberg, H., 1996. Vegetation Mitteleuropas mit den Alpen in         pp. 252.
  ¨
  okologischer, dynamischer und historischer Sicht, 5. Aufl Ulmer                 ¨      ¨      ¨   ¨
                                     Isermann, M., 2001. Kornungsverhaltnisse in Kustendunen Meck-
  Verlag, Stuttgart, pp. 511–561.                      lenburg-Vorpommerns. Proceedings Forschung und Naturschutz
                 ¨
Ellenberg, H., Weber, H.E., Dull, R., Wirth, V., Werner, W., Pau-                ¨
                                        in Sandlebensraumen. Freiburg 15.–17.02.2001, pp. 61–
  lißen, D., 1992. Zeigerwerte von Pflanzen in Mitteleuropa.         64.
  Scripta Geobot. XVIII, 258.                      Isermann, M., 1997. Vegetations- und standortskundliche Untersu-
   ´
Garcıa-Mora, M.R., Gallego-Fernandez, J.B., Williams, A.T., Gar-               ¨   ¨
                                        chungen in Kustendunen Vorpommerns. PhD Thesis, University
   ´
  cıa-Novo, F., 2001. A coastal dune vulnerability classification. A     of Greifswald, 323 pp.
  case study of the SW Iberian Peninsula. J. Coast. Res. 17 (4),    Isermann, M., 1995. Dunes in contradiction with different interests.
  802–811.                                  An example: the camping-ground prerow (Darß/Baltic Sea). In:
   ´               ´
Garcıa-Mora, M.R., Gallego-Fernandez, J.B., Williams, A.T., Gar-        Salman, A.H.P.M., Bonazountas, M. (Eds.), Coastal Manage-
  cia-Novo, F., 2000. Plant diversity as a suitable tool for coastal     ment and Habitat Conservation, vol. I. Proceedings of the Fourth
  dune vulnerability assessment. J. Coast. Res. 16 (4), 990–995.       EUCC Conference in Marathon, Greece. Publ, EUCC, pp. 439–
   ´        ´
Garcıa Novo, F., Diaz Barradas, M.C., Zunzunegui, M., Garcia    ´     449.
               ´
  Mora, R., Gallego Fernandez, J.B., 2004. Plant functional types                        ¨
                                     Jeschke, L., 1985. Vegetationsveranderungen in den Kusten-   ¨
                    ´
  in coastal dune habitats. In: Martınez, M.L., Psuty, N.P. (Eds.),     landschaften durch Massentourismus und Nutzungsintensivier-
  Coastal Dunes—Ecology and Conservation. Ecological Stu-          ung. Archiv Naturschutz und Landschaftsforschung 25, 223–
  dies., vol. 171. Springer, pp. 155–172.                  236.
Goldberg, D.E., Estabrook, G.F., 1998. Separating the effects of     Kooijman, A., Besse, M., 2002. The higher availability of N and P in
  numbers of individuals sampled and competition on species         lime-poor than in lime-rich coastal dunes in the Netherlands. J.
  diversity: an experimental and analytical approach. J. Ecol. 86,      Ecol. 90, 394–403.
  983–988.                                             ¨         ¨
                                     Kowarik, I., 1999. Naturlichkeit, Naturnahe und Hemerobie als
Grime, J.P., 2002. Plant Strategies, Vegetation Processes, and Eco-                           ¨
                                        Bewertungskriterien. In: Konold, W., Bocker, R., Hampicke,
  system Properties, 2nd ed. Chichester, New York Weinheim,                      ¨
                                        U. (Eds.), Handbuch fur Naturschutz und Landschaftspflege.
  Brisbane, Singapore, Toronto, 417 pp.                   pp. 1–18.
                   ¨
Grunewald, R., 2004a. Biodiversitat und Tourismus: Auswirkungen      Knapp, H.D., 1996. Belastung und Schutz der Boddenlandschaften
                          ¨
  des Fremdenverkehrs auf die Strand- und Dunenvegetation der                      ¨
                                        an der deuteschen Ostseekuste. In: Buchwald, K., Engelhardt,
                      ¨
  Pommerschen Bucht. Bundesamt fur Naturschutz: Treffpunkt          W. (Eds.), Umweltschutz—Grundlagen und Praxis, Band 6/II.
  Biologische Vielfalt 4, 129–136.                      Economica Verlag, Bonn, pp. 76–112.
Grunewald, R., 2004b. Auswirkungen des Fremdenverkehrs auf die      Magurran, A.E., 1988. Ecological Diversity and its Measurements.
    ¨      ¨            ¨
  Dunen- und Spulsaumvegetation auf Rugen und Usedom unter          Princeton University Press, Princeton, pp. 179.
           ¨                   ¨
  besonderer Berucksichtigung verschiedener Diversitatsindizes     Martinez, M.L., Psuty, N.P. (Eds.), 2004. Coastal Dunes—Ecology
  (Effects of recreational activities on the dune and beach vegeta-     and Conservation. Ecological Studies, 171. Springer, pp.386.
  tion on Ruegen and Usedom Island). Rostock. Meeresbiolog.       Maun, M.A., 2004. Burial of plants as a selective force in sand
  Beitr. 12, 71–88.                             dunes. In: Martinez, M.L., Psuty, N.P. (Eds.), Coastal Dunes—
Grunewald, R., Łabuz, T.A., 2004. Plant diversity on dunes of Swina      Ecology and Conservation. Ecological Studies, 171. pp. 119–
  Gate Barrier: a largely undisturbed accumulative coast. Coast.       135.
  Rep. 2, 139–147.                           Mouillot, D., Sylvain, G., Aliaume, C., Verlaque, M., Belsher, T.,
Haeupler, H., Muer, T., 2000. Bildatlas der Farn- und Bluetenp-        Troussellier, M., Do Chi Thang, 2005. Ability of taxonomic
  flanzen Deutschlands. Ulmer Verlag, Stuttgart, pp. 759.           diversity indices to discriminate coastal lagoon environments
Haeupler, H., 1982. Evenness als Ausdruck der Vielfalt in der         based on macrophyte communities. Ecol. Ind. 5 (1), 1–71.
  Vegetation. J. Cramer, Vaduz, 260 pp.                           ˆ
                                     Mouillot, D., Lepretre, A., 1999. A comparison of species estima-
Hellemaa, P., 1998. The Development of Coastal Dunes and their         tors. Res. Popul. Ecol. 41, 203–215.
  Vegetation in Finland, 176. University of Helsinki, Faculty of      ¨             ¨
                                     Muller, D., 2004. Naturraumliche Grundlagen Mecklenburg-Vor-
  Science, Department of Geography, Fennia, pp. 157.             pommern. In: Berg, C., Dengler, J., Abdank, A., Isermann, M.
Hesp, P.A., 1991. Ecological processes and plant adaptations on        (Eds.), Die Pflanzengesellschaften Mecklenburg-Vorpommerns
  coastal dunes. J. Arid Environ. 21, 165–191.                       ¨
                                        und ihre Gefahrdung. Weissdorn-Verlag, Jena, pp. 22–26.
                    R. Grunewald, H. Schubert / Ecological Indicators 7 (2007) 1–21                     21

Packham, J.R., Willis, A.J., 1997. Ecology of dunes. Salt Marsh and    Reinhard, H., 1962. Klimatologie. In: VEB – Topographischer
  Shingle 335.                               Dienst Schwerin, I. Natur des Landes. Atlas f. d. Bezirke
¨
Partel, M., Zobel, M., Zobel, K., van der Maarel, E., 1996. The        Rostock, Schwerin und Neubrandenburg.
  species pool and ist relation to species richness: evidence from   Rodgers, J.C., Parker, K.C., 2003. Distribution of alien plant species
  Estonian plant communities. OIKOS 75, 111–117.              in relation to human disturbance on the Georgia Sea Islands.
Patil, G.P., Taillie, C., 1982. Diversity as a concept and its measure-    Divers. Distribut. 9 (5), 385–398.
  ment. J. Am. Statist. Assoc. 77/379, 548–561.             Rodgers, J.C., 2002. Effects of human disturbance on the dune
Peet, R.E., 1974. The measurement of species diversity. Ann. Rev.       vegetation of the Georgia Sea Islands. Phys. Geogr. 23 (1), 79–
  Ecol. Syst. 5, 285–307.                          94.
Pickett, S.T.A., White, P.S. (Eds.), 1985. The Ecology of Natural     Seidel, W., 2001. Tourismus in Mecklenburg-Vorpommern 1999.
  Disturbance and Patch Dynamics. Academic Press, Orlando,         Statistische Sonderhefte des Statistischen Landesamtes Meck-
  Florida, pp. 472.                             lenburg-Vorpommern Jg. 11, H3, 59 pp.
Pielou, E.C., 1966. The measurement of diversity in different types    Shannon, C.E., Weaver, W., 1949. A Mathematical Theory of
  of biological collections. J. Theoret. Biol. 13, 131–144.         Communication. University of Illinois Press, pp. 125.
Piotrowska, H., Gros, K., 1998. Natural and anthropogenic changes     Tilman, D., Knops, J., Wedin, D., Reich, P., 2001. Experimental and
  in sanddunes and their vegetaton on the Southern Baltic coast.      observational studies of diversity, productivity, and stability. In:
  Perspect. Coast. Dune Manage. 33–40.                   Kinzig, P., Pagala, S.W., Tilman, D. (Eds.), The Functional
Piotrowska, H., 1988. The dynamics of the dune vegetation on the       Consequences of Biodiversity. Princeton University Press,
  Polish Baltic Coast. Vegetation 77, 169–175.               pp. 42–70.
Piotrowska, H., 1979. Natural resistance against mechanical        Townsend, C.R., Harper, J.L., Begon, M., 2002. Essentials of
  destruction in plant communities on Baltic Coast dunes. Bio-       Ecology. Blackwell Publishers, Oxford, pp. 647.
  logia 1, 5–13.                            v. Nordheim, H., Boedeker, D. (Eds.). 1998. Red List of Biotopes and
Reinhard, H., 1951. Karten zur Klimabewertung bei der Bod-          Biotope Complexes of the Baltic Sea, Belt Sea and Kattegat. In:
     ¨
  enschatzung in Mecklenburg. Angew. Meteorol. 1 (1).            Baltic Sea Environment Proceedings (Helcom), vol. 75, 115 pp.