© 2010 Gebrüder Borntraeger, 70176 Stuttgart, Germany www.borntraeger-cramer.deDOI: 10.1127/0340 – 269X/2010/0040– 0426 0340-269X/10/0040-0426 $ 06.30

Phytocoenologia, 40 (1), 1–14Stuttgart, May 17, 2010

Vegetation patterns and primary succession on sea-born volcanic islands

(Santorini archipelago, Aegean Sea, Greece)

by Panayotis DIMOPOULOS, Agrinio, Greece, Thomas RAUS, Berlin, Germany, Ladislav MUCINA, Perth,

Australia and Ioannis TSIRIPIDIS, Thessaloniki, Greece

with 4 fi gures and 5 tables

Abstract. Santorini (S Cyclades, Greece) is one of the active volcanoes of the South Aegean volcanic arc, with more than 12 major eruptions during the past 250,000 years. Our research, using a set of 93 phytosociological relevés, focused on the two volcanic islands of Palea Kameni (PK) and Nea Kameni (NK), differing in age and history of vegetation development. Numerical classifi cation techniques were used to defi ne plant communities, and Redun-dancy Analysis (RDA) was used to investigate the main environmental gradients and identify the explanatory vari-ables with a unique signifi cant contribution to the explanation of the communities’ fl oristic composition. In RDA environmental factors such as altitude, slope inclination, relief, aspect, age of geological substrate, distance from the shore, and selected Ellenberg Indicator Values (EIVs) were used. In order to test the robustness of the RDA results, principal components analysis was applied to the relevés data set as well. Partial RDAs (variance partitioning) were also applied to investigated the proportion of variance attributed to the age of the lava substrate as compared to the rest of environmental factors. The classifi cation revealed seven plant communities (four on PK and three on NK), described in terms of unique species composition, characteristic ecology, and distribution. Three of the communities were dominated by perennials (all found on PK), while the other communities were built prevalently of therophytes (all communities found on NK and one occurring on PK). The age of lava substrate was found among the most important and statistically signifi cant variables. Furthermore, EIVs for light, nitrogen, reaction, and salinity were found to explain relatively high and statistically signifi cant proportions of species data variance. However, the pro-portion of variance attributed purely to the age of lava substrate was by far smaller than that attributed to the rest explanatory variables. Conclusively, we suggest that the formation of local pioneer vegetation on the studied volca-nic islands is not solely controlled by the age of the substrate (“maturation” of the communities through saturation of their local species pools), but it is also under strong control of factors such as the distribution of ashes after the recent volcanic eruptions, and the chemical and physical properties (e.g. nutrient content) of the soils formed by the deposition of the ashes.

Abbreviations: Comm. – Community, EIVs – Ellenberg Indicator Values, NK – Nea Kameni, PCA – Principal Components Analysis, PK – Palea Kameni, RDA – Redundancy analysis.

Keywords: halo-nitrophilous scrub, Mediterranean vegetation, ordination, tephra, therophytic vegetation, vegeta-tion dynamics, Kameni islands, volcano.

Nomenclature: GREUTER et al. (1984–1989), RAUS (1991), TUTIN et al. (1968–1980, 1993).

Introduction

Santorini (S Cyclades, Greece) is one of the active volcanoes of the South Aegean Volcanic Arc (known also as Hellenic Volcanic Arc; ALBANAKIS & STYLLAS 2004); it produced more than 12 major eruptions within the past 250,000 years. The major Santorini (or Thera as it was called in classic Hellenic times) erup-tion, sometime during a period ranging from 1627 to 1600 B.C. (FRIEDRICH et al. 2006) and associated tsunamis are linked to the demise of Minoan Civiliza-tion of the Cyclades and Crete (see BOND & SPARKS 1976, KELLER 1980, DOWNEY & TARLING 1984, SPARKS 2000 etc.). Currently, the Santorini volcano is

in a state of interruptive repose, but there is no reason to believe that there will not be any future eruptions (PYLE & ELLIOTT 2006). According to PAPAZACHOS (1990) the mean repeat time of the strong eruptions in the Santorini volcanic centre during the last fi ve centuries is 94 (±43) years, resulting in probability of 0.6 for eruption to occur within the next 50 years and probability of 0.95 for the next 110 years (for fore-casting the next eruption see FYTIKAS et al. 1990 and PAPADOPOULOS & ORFANOGIANNAKI 2005). PYLE & ELLIOTT (2006) suggested for instance, that if an eruption should have occurred in 2006 it would have lasted for more than 2.7 years, and eventually pro-duced a dome 115–125 m high. Because of the poten-

2 P. Dimopoulos et al.

tial security risks to property, health and life of both the archipelago’s inhabitants and the large number of tourists regularly visiting Santorini, the site has been monitored by the European Commission as one of the fi ve European Laboratory Volcanoes (CASALE et al. 1998).

Inside the seawater-fi lled Santorini caldera, sea-born, uninhabited volcanic islands of Palea Kameni (PK) and Nea Kameni (NK) offer the unique oppor-tunity of assessing and monitoring primary vegeta-tion succession. The 2200-years old PK is currently covered by a mosaic of evergreen sclerophyllous and semi-halophytic shrub communities as well as short-lived, seasonal therophytic herb-rich swards and grasslands. Apart from three individuals of woody Ficus carica which survived 20th century’s volcanic eruptions, the most evolved vegetation type on NK, which was formed in ca. 1570–1950 A.D. by four subsequent eruptive phases of the Santorini volcano, is an open-canopy pioneer community rich in grasses and legumes (RAUS 1988). The fi rst and only relevés of actual vegetation units on NK are published by RAUS (1986: 388 & 391), while published vegetation data from PK were then still lacking. That is why a renewed study on the vegetation on the Kameni is-lands is postulated by MAKRIS (2006) to assess the lo-cal vegetation dynamics within the last 20 years.

The main objective of the present paper is to pro-vide baseline data on current vegetation of PK and NK in the Santorini Archipelago. In particular we aim (a) to describe major vegetation patterns (plant communities), (b) to identify the leading ecological factors controlling the vegetation patterning on both islands, and (c) to discuss the pathways of the pri-mary vegetation succession on the studied islands.

We believe that our baseline descriptive study will yield data on the status of the local biodiversity which then become valuable in assessments of the extent of damage caused by an eruption or eruptions, and would assist in judgments on the patterns and path-ways of vegetation development (and/or recovery) after optional almost unavoidable future eruptive events in the area (for details on local seismic activity in 1994–2002, see DIMITRIADIS et al. 2005).

Materials and Methods

Study area: Topography and environmental factors

The Santorini Archipelago (36 º19’56” to 36 º28’40” N and 25 º19’22” to 25 º29’13” E) is the southern-most island group of the Cyclades, situated in the Aegean Sea, 260 km SE of Athens and 120 km (63 nautical miles) N of the island of Crete. Origin and present confi guration of its fi ve islands (Fig. 1) are the result of repeated volcanic activities (for details see DRUITT et al. 1999, MCBIRNEY 2000, FRIEDRICH 2004). Around 1613 (i.e. 1627–1600) B.C. the pre-Kameni volcano exploded in what is today known

as Minoan Eruption, transforming the then existing caldera (FRIEDRICH et al. 1988, 2000, 2006, DRUITT & FRANCAVIGLIA 1992) to its present extent. Addi-tional collapsing of the caldera wall left three islands forming a broken ring including the islands of Thira (75.8 km², 568 m alt., 12.440 inhabitants in 2001), Thi-rasia (9.3 km², 295 m alt., 268 inhabitants in 2001) and Aspronisi (0.13 km², 71 m alt., uninhabited). Roughly 1420 years after the Minoan Eruption, a magmatic refi lling of the caldera began. In the central part of the caldera, sea-born islands called the Kameni Is-lands (orthography follows PYLE & ELLIOTT 2006; ‘Kaimeni’ Islands of CAMMERLOHER 1935, PICHLER & KUSSMAUL 1980 and RAUS 1986, 1988; elsewhere as in DIAPOULIS 1971 given as ‘Kammeni’ Islands, i.e. the burnt islands) emerged and re-formed in a series of later eruptions. Today, these two islands, such as Palea Kameni (meaning “Old”; PK) and Nea Kameni (meaning “New”; NK; see maps in PYLE & ELLIOTT 2006: 254, Fig. 1 & 258–261, Fig. 4–8) represent the emergent top of a 2.5 km³ volcano that has a basal area of ca. 3.5 km² and that rises 500 m from the fl oor of the fl ooded caldera (DRUITT et al. 1989, 1999). PK emerged in 46–47 A.D. and today it is covering an area of 0.54 km² after having been enlarged by ascending submarine lava in 726 and after partial collapse ex-perienced in 1457 (FOUQUÉ 1879, MCBIRNEY 1999). The island reaches the highest altitude of 103 m a.s.l. NK received its present shape through eight eruptive phases between ca. 1570 and 1950. It is thus more than 1500 years younger than PK and came into existence as a single island (about 2 km diameter) only after 17 August 1925, when the two former islands of Mikra Kameni (emerged around 1570) and Old Nea Ka-meni (emerged from 1707 onwards) were united by a large lava-fl ow (WASHINGTON 1926, RECK 1936). The last eruption of NK took place in 1950 (GEORGALAS 1953), covering a part of the older lava surfaces with fresh ashes. At present, NK exhibits a surface area of 3.44 km² and it is reaching 124 m of altitude (for the age of the different lava surfaces on NK, see PYLE & ELLIOTT 2006: 260, map 7). Only lithosols, at most showing immature A-C profi les, are developed on both islands. In general, raw lava-blocks with crev-ices, and coarse to fi ne ashes, serve as substrates for higher plants. There is a complete lack of permanent surface water and freshwater springs. A consider-able part of the annual rainfall is lost by runoff. The macroclimate is of a low-precipitation, dry Medi-terranean (winter-rainfall, summer-dry) type, with a mean temperature of about 25 ° C in the hottest, and of 11 ° C in the coldest month; there is generally no frost in winter. Mean annual precipitation attains only 357 mm, with the dry season lasting from April to October (SCHMALFUSS et al. 1981). Frequent over-heating of mostly black-coloured tainted volcanic substrates is probably the most outstanding feature of the local microclimate, enhancing the stress levels (due to water defi ciency) for all sessile organisms, in particular for juvenile plant seedlings (e.g. RICHTER 1984). In addition, steady winds in summer enhance

Vegetation patterns and primary succession on sea-born volcanic islands (Santorini archipelago, Aegean Sea, Greece) 3

evaporation. Both islands are uninhabited. Grazing occurred irregularly and intermittently on PK in the past (the shepherd was then single, temporary inhab-itant; local comm. and MAKRIS 2006). On the unin-habited NK, some active fumaroles attract thousands of tourists every year who enter the island through a small harbour on its northern coast and walk to the island’s top. A second harbour on the eastern coast of NK, and the single harbour site at the northern coast of PK, are frequented almost exclusively by lo-cal fi shermen and researchers.

Vegetation data: Past and present

The fi rst (partial) survey on the vascular plants of the Kameni Islands took place in April 1911 when Profes-sor Vierhapper visited the islands with a group of stu-dents of the University of Vienna, Austria. This group made a fl oristic inventory of NK and presented some hints on supra-littoral and inland vegetation types (VIERHAPPER 1914, 1919). Based on this milestone of botanical research in the area, CAMMERLOHER (1935) repeated the NK fl oristic inventory and described the impact of the eruptions of 1925–1928 on the plant cover of NK. The subsequent recovery of the vascular plant diversity on NK (although infl uenced to an un-known extent by eruptions in 1939–1941 and in 1950) was documented in April 1967 by HANSEN (1971) who also contributed the fi rst preliminary notes on

the fl ora of PK as well. All the above-mentioned au-thors more or less exclusively dealt with fl ora statis-tics rather than with primary vegetation development from a phytosociological point of view.

17 years after Hansen’s fi eldwork in the area, one of us (T. Raus) made the fi rst Braun-Blanquet stan-dard vegetation relevés on both NK and PK with an aim to document the formation of ‘unsaturated’ plant communities in the course of local primary succes-sion on volcanic surfaces. The NK data were pub-lished (RAUS 1986: pp. 388, 391), while the PK data remained unpublished, however are now being used in our analyses. In April 2000, two other of the au-thors (P. Dimopoulos & L. Mucina) carried out ad-ditional vegetation sampling on both islands in order to complement the 1984 survey (RAUS 1986). Both available relevé data sets add up to 93 samples and form the basis of our current study. Depending on the vegetation type, size of plots varies from 6 m² (e.g. lava crevices) to 100 m² (e.g. extended tephra slopes). BRAUN-BLANQUET (1964) sampling scale was used by RAUS (1986), while we used modifi ed (see BARKMAN et al. 1964) Braun-Blanquet sampling scale in our 2000 survey.

Environmental data

The environmental data set included physio-geo-graphic parameters: altitude (in m), inclination (in

Fig. 1. Map of the present-day confi guration of the islands of Santorini Archipelago, as a result of repeated volcanic activities. The studied Kameni islands (NK and PK) are located in the central part of the caldera.

4 P. Dimopoulos et al.

degree), relief (for states see below), aspect (trans-formed as described below), geographic coordinates, as well as the Ellenberg Indicator Values (EIVs). Three different types of relief were distinguished: plane, slope and ridge. In the numerical analyses, re-lief was modifi ed to a quasi-ordinal scale, using the values 1, 2 and 3 for the plane, slope and ridge type of relief, respectively, representing a surrogating de-creasing soil depth. Aspect (measured in the fi eld in degree as deviation from the North) was transformed through calculation of potential annual direct inci-dent radiation and heat load, using the procedures proposed by MCCUNE & KEON (2002). With the help of the geographical coordinates of the relevés and topographical as well as geological maps taken from PYLE & ELLIOTT (2006), the age of the geological sub-strate and the distance from the shore (coast) for each relevé were established. The EIVs for relevés were calculated as weighted averages (by relative cover) of the EIVs of species occurring in the relevé. The EIVs for species were adopted from BÖHLING et al. (2002), with few new additions and minor modifi cations where considered necessary. Species relative cover was calculated on the basis of the power (value: 0.3) transformed values of the cover percentages. In this way the cover values have been taken into account in the determination of the relevé EIVs, but the latter were not governed by the high cover values of the dominant species.

Data analyses

Species and environmental data were stored and managed by a Turboveg (HENNEKENS & SCHAMINÉE 2001) database. Braun-Blanquet cover codes were re-placed by percentage cover values as follows r: 1 %, +: 2 %, 1: 3 %, 2: 13 %, 2 m: 4 %, 2a: 8 %, 2b: 18 %, 3: 38 %, 4: 68 %, and 5: 88 %. The classifi cation of the relevés was done using a cluster analysis, involv-ing fl exible beta linkage method (with b set equal to –0.25) and using Sørensen (Bray-Curtis) resemblance (see PODANI 1994 and LEGENDRE & LEGENDRE 1998 for details on the properties of the parameters cho-sen). Initially a relatively large number of clusters have been distinguished. The fi nal pattern of clusters (communities) was decided upon based on presence of diagnostic species and the relative position of the relevés on the ordination diagrams. The preliminary assignment of the detected communities to classes follows MUCINA (1997). As diagnostic were consid-ered the species with differentiating frequency and/or cover in one or a combination of communities. De-termination of diagnostic species was assisted by the algorithm by TSIRIPIDIS et al. (2009); species cover differentiation was also considered.

Ordination analyses were applied to explore the infl uence of environmental variables on the vegeta-tion patterns. The direct ordination method of RDA was applied in order to test the signifi cance of the re-lationship of the explanatory variables with the spe-

cies composition of the relevés. Redundancy Analy-sis (RDA) was based on two sub-sets of explanatory variables. Initially, seven variables were used, namely the lava age, altitude, ground slope, relief, distance from the shore, potential annual direct incident ra-diation, and heat load. After evaluating the results of the analyses six more variables were added, such as the EIVs for light, temperature, moisture, reaction, nitrogen, and salinity. The results obtained using the two different data sets of explanatory variables were compared between each other using as criteria the distribution of relevés in diagrams, the correlation of explanatory variables with the RDA axes as well as the total variance explained. Principal components analysis (PCA) was applied in order to obtain insight into vegetation patterns based on fl oristic data only and to test robustness of the direct gradient analyses. By comparing the results of PCA and RDA, we es-timated if the explanatory variables used in the latter analysis are adequate to reveal the main gradients in the data set. We choose to embark on this compari-son being aware of the fact that the use of EIVs in a RDA may implicate a circularity argument, as these explanatory variables are derived from the fl oristic composition of the relevés. The ordination analyses were applied in the data set of all relevés as well as in a partial data set resulted after the removal of the most distinguishable communities. PCA and RDA were applied using the covariance matrix. In RDA a manual forward selection of the statistically sig-nifi cant variables was applied using Monte Carlo test with 999 permutations.

In both PCA and RDA species covers were trans-formed according to chord transformation proposed by LEGENDRE & GALLAGHER (2001). After this trans-formation, the Euclidean distances are equal to the chord distances, which are the distances (chords) joining two points (sites) within a segment of a hyper-sphere of radius 1 (LEGENDRE & GALLAGHER 2001). This transformation is known in some texts as “nor-malization” (e.g. PODANI 1994). The chord transfor-mation is an appropriate solution in order to avoid the high weights given to rare species and to species rich relevés in the case of the PCA and RDA applications if the data matrix contains many zeros or when the species richness values (per relevé) vary considerably. LEGENDRE & GALLAGHER (2001) suggested that the use of such transformation was appropriate especially in instances when investigation of the signifi cance of the relationships between species composition and a set of explanatory variables was intended.

Partial RDA was also applied in order to test for the effect of lava age (as opposed to other “environmen-tal” variables) on the fl oristic composition of the plant communities. The variance partitioning (ZUUR et al. 2007) was applied in the same relevés data sets where RDAs were made as well as in a data set comprised of the relevés of NK. Partial RDAs were done using the 13 variables explanatory data set. In order to avoid an over-explanation of variance, fi rstly a RDA was run using all the explanatory variables except lava age and

Vegetation patterns and primary succession on sea-born volcanic islands (Santorini archipelago, Aegean Sea, Greece) 5

those presenting a signifi cant contribution (tested by means of Monte Carlo test with 999 permutations) to the explanation of variance were used in variance par-titioning (in the description below these variables are called signifi cant variables). Then fi ve RDAs were ap-plied having the following explanatory variables and covariables: 1) lava age and the signifi cant variables as explanatory variables, 2) lava age as explanatory variable, 3) the signifi cant variables as explanatory variables, 4) lava age as explanatory variable and the signifi cant variables as covariables and 5) the signifi -cant variables as explanatory variables and lava age as covariable. Then following the calculations described in ZUUR et al. (2007) the proportion of explained variance attributed purely to the age of lava substrate and the other signifi cant (physiographic and EIVs) variables, as well as the shared effects and the residual variance was found.

In both classifi cation and ordination analyses spe-cies occurring in one or two relevés were omitted from the numerical analyses. After testing and inter-pretation of the results it was decided not to apply any transformation on species cover values in the classifi cation, while the chord transformation for PCA and RDA was done on the square-root trans-formed species cover values.

Classifi cation was performed using PC-ORD 5 (MCCUNE & MEFFORD 1999, MCCUNE & GRACE 2002) and JUICE 6.5 (TICHÝ 2002, TICHÝ & HOLT 2006), while ordination analyses were done using CANOCO 4.5 software package (TER BRAAK & ŠMILAUER 2002).

Results

Descriptions of plant communities

Seven plant communities, each characterised by dis-tinct fl oristic composition, set of diagnostic species as well as structure and ecological traits, were dis-tinguished. The characteristic features of the com-munities, as well as their preliminary assignment to vegetation classes are given below; the order of com-munities follows the sequence of columns in the syn-optic Table 1.

Comm. 1: Limonium graecum (Crithmo-Staticetea)

Distribution: PK (5 relevés, all sampled in 2000).

Diagnostic species and structure: loose perennial matrix made up by the suffruticose halophyte Limo-nium graecum accompanied by a bulk of annuals of which Sedum litoreum shows differential focus of oc-currence; no larger shrubs able to compete because of frequent seawater inundation.

Ecology: Aged coastal rock, crevices and ledges, heavily salt-sprayed.

Comm. 2: Atriplex halimus–Lycium schweinfurthii (Pegano-Salsoletea) (Fig. 2)

Distribution: PK, particularly S- and SW coast (6 relevés, all sampled in 1984).

Diagnostic species and structure: loose to locally dense and intricate thickets of halophytic shrubs of Atriplex halimus and Lycium schweinfurthii, with un-dergrowing Limonium graecum weakened by com-petition, accompanied by a bulk of annuals of which Bellium minutum, Malcolmia fl exuosa, Mesembryan-themum nodifl orum, Sedum litoreum, Silene sedoides, Sonchus oleraceus show differential halotolerant indi-cator value.

Ecology: Halo-nitrophilous scrub on aged coastal rocks and debris, salt-sprayed.

Comm. 3: Pistacia lentiscus (Quercetea i l icis)

Distribution: PK (8 relevés, 6 sampled in 2000 and 2 in 1984).

Diagnostic species and structure: Dense, fl at, wind-shaped thickets of woody Pistacia lentiscus with procumbent-growing trunks and branches, with the hemicryptophytic grass Dactylis glomerata subsp. hispanica and the geophytic composite Aetheorhiza bulbosa subsp. microcephala showing differential fo-cus of occurrence.

Ecology: Non-halophytic stands occurring on nar-row ledges of steep coastal cliffs beyond the surge; the most prominent vegetation unit on the island’s plateau on aged deep ashes and rocky debris.

Comm. 4: Lupinus angustifolius–Helichrysum italicum (Thero-Brachypodietea ramosi) (Fig. 3)

Distribution: NK (30 relevés, 23 sampled in 2000 and 7 in 1984).

Diagnostic species and structure: permanent ma-trix of the semi-shrub Helichrysum italicum growing spaced because of underground root competition, accompanied by a bulk of annuals of which Lupinus angustifolius excels by biomass productivity.

Ecology: Non-halophytic, therophytic vegetation on young, deep, fi ne ashes.

Comm. 5: Lupinus angustifolius–Hyparrhenia hirta (Thero-Brachypodietea ramosi)

Distribution: NK (16 relevés, all sampled in 1984).

6 P. Dimopoulos et al.

Table 1. Synoptic table of constancy values. Species median cover values are presented as exponents to constancy values (% median cover values have been replaced by codes of Braun-Blanquet scale). The diagnostic taxa of the Communities (1 through 7, as described in the text) are presented in bold. Con: pooled species constancy (for the entire table).

Community Con 1 2 3 4 5 6 7

No. of relevés 5 6 8 30 16 17 11

Sedum litoreum 27 80+ 1001 38+ 23+ 38+ . 9+

Limonium graecum 5 401 50+ . . . . .

Atriplex halimus 6 . 1003 . . . . .

Bellium minutum 6 . 1001 . . . . .

Malcolmia fl exuosa subsp. fl exuosa

10 40+ 100+ 25+ . . . .

Lycium schweinfurthii 6 . 832 13+ . . . .

Sonchus oleraceus 6 . 831 . 3r . . .

Silene sedoides 3 . 501 . . . . .

Mesembryanthemum nodifl orum

4 20r 501 . . . . .

Pistacia lentiscus 8 . . 1005 . . . .

Dactylis glomerata subsp. hispanica

9 . . 88+ . . . 18+

Aetheorhiza bulbosa subsp. microcephala

7 20+ . 631 3+ . . .

Helichrysum italicum 31 . . . 671 56+ 12+ .

Lupinus angustifolius 51 . . . 60+ 1003 71+ 45+

Hyparrhenia hirta 36 . . 13+ 172 1003 822 .

Umbilicus rupestris 24 20r 33+ 38+ 33+ . 41+ 9r

Andryala integrifolia 50 . 50+ . 27+ 881 82+ 1002

Psilurus incurvus 42 40+ . 131 631 . 531 1002

Tolpis barbata 35 20r . 13+ 37+ 13+ 53+ 1002

Hypochaeris glabra 46 60+ . 13+ 20r 69+ 82+ 1001

Crassula alata 28 . . . 30+ 6+ 472 911

Anagallis arvensis 12 . . 25+ . . . 911

Ononis reclinata 11 20r . . . . . 911

Asterolinon linum-stellatum

9 . . 13+ . . . 731

Anthemis tomentosa 7 . . . . . . 64+

Ononis diffusa 6 . 171 . . . . 451

Centaurium tenuifl o-rum

4 . . . . . . 36+

Silene gallica 5 . . . 3+ . . 36+

Crepis foetida 3 . . . . . . 27+

Trifolium suffocatum 4 . . . . . 6+ 27+

Rumex bucephalopho-rus subsp. aegaeus

82 100+ 83+ 38+ 1002 75+ 94+ 1001

Lagurus ovatus 55 60+ 67+ 38+ 37+ 1002 651 64+

Vulpia ciliata 54 . . . 47+ 88+ 882 1002

Corynephorus articu-latus

53 . . . 631 81+ 65+ 911

Crepis multifl ora 51 60+ 100+ 50+ 20+ 88+ 591 73+

Filago gallica 50 40r . 13+ 73+ 6+ 821 911

Polycarpon tetraphyl-lum

44 20+ . 13+ 43+ 75+ 82+ 27+

Trifolium arvense 41 . 67+ 13+ 23+ 44+ 942 55+

Avena barbata 41 40+ . 25+ 23+ 1002 41+ 64+

Community Con 1 2 3 4 5 6 7

No. of relevés 5 6 8 30 16 17 11

Trifolium campestre 39 . . 25+ 20+ 63+ 652 911

Lamarckia aurea 31 601 83+ 25+ 10+ 44+ 531 18+

Urospermum picroides 31 40r 50+ 38r 13r 751 29+ 18r

Galium recurvum 29 80+ . 50+ 40+ . 35+ 27+

Parietaria cretica 29 80+ . 13+ 20+ 44+ 65+ .

Rostraria cristata 28 40+ 1001 13+ 7+ . 59+ 641

Sagina apetala 20 20+ . . 27+ . 351 45+

Erodium cicutarium 19 100+ . 13+ 3r . 47+ 36+

Geranium molle subsp. molle

19 60+ 50+ 38r 17r . 24+ 9r

Senecio vulgaris 18 . . 25+ 30+ . 18+ 36+

Bromus madritensis subsp. madritensis

17 802 50+ 381 . . 411 .

Cerastium semidecan-drum

17 . . . 57+ . . .

Senecio gallicus 17 . 67+ . 20+ 19+ 24+ .

Plantago lagopus 15 20+ 50+ 13+ . . . 912

Biscutella didyma 15 40+ . 38+ 13r . 29+ 9r

Euphorbia peplus 14 40+ . 50+ . . 241 36+

Ornithopus compressus 14 . . . . 13+ 291 64+

Trifolium scabrum 13 . 672 . . . 411 18+

Ornithopus pinnatus 13 . . . . . 412 551

Echium arenarium 12 20+ 83+ 13r . . . 45+

Clypeola jonthlaspi 12 80+ . 25+ 13+ . . 18+

Vicia lathyroides 12 40r . 50r 3+ . 18+ 18+

Trifolium glomeratum 12 . . . . 13+ 351 362

Linaria pelisseriana 12 . . . 13+ 6+ 24+ 271

Plantago weldenii 11 40r 50+ 13+ . . . 45+

Stipa capensis 11 . 50+ . . 6+ . 642

Bromus tectorum 11 20+ 50+ . 10+ 13+ 61 9+

Hyoseris scabra 10 80+ . 38+ . . . 27+

Bromus intermedius 10 . . 131 3+ 38+ . 18+

Brachypodium dis-tachyon

9 20+ 501 251 3+ 6+ 6+ .

Sonchus tenerrimus 9 . . 13+ . 25+ 24r .

Astragalus pelecinus 8 . 172 . . . . 642

Cuscuta palaestina 8 . . 13+ 3+ . . 551

Aira cupaniana 8 . . . 231 . . 91

Anogramma lepto-phylla

8 . . . 23+ . 6+ .

Carlina graeca 8 . . 13+ . 38+ . 9r

Hordeum leporinum 7 602 33+ 131 . . . 9+

Bromus fasciculatus 7 40+ . 132 3+ . 6+ 181

Carduus pycnocephalus 7 40+ . 13r . 6+ 18+ .

Misopates orontium 7 . . . . 19+ 24+ .

Valantia hispida 7 . . 13+ 17+ 6+ . .

Parentucellia latifolia 6 . 1002 . . . . .

Vegetation patterns and primary succession on sea-born volcanic islands (Santorini archipelago, Aegean Sea, Greece) 7

Diagnostic species and structure: Usually dense to very dense stands of the perennial grass Hyparrhe-nia hirta and the annual grasses Avena barbata and Lagurus ovatus, accompanied by a bulk of non-grassy annuals of which Lupinus angustifolius excels by bio-mass productivity and optionally rivals Hyparrhenia in cover-value, and with the composites Andryala in-tegrifolia, Crepis multifl ora and Urospermum picroi-des showing differential focus of occurrence; the most closed vegetation unit of NK.

Ecology: Non-halophytic, non-woody, steppe-like vegetation on young shallow fi ne ashes.

Comm. 6: Lupinus angustifolius–Umbilicus rupestris (Helianthemetea guttati)

Distribution: NK (17 relevés, 9 sampled in 2000 and 8 in 1984).

Diagnostic species and structure: Floristically simi-lar to the preceding unit but less dense, particularly differing in Umbilicus rupestris showing differential focus of occurrence as an indicator of open, rocky habitats.

Ecology: Young coarse ashes rich in lava rocks caus-ing additional “runoff” water from dew.

Community Con 1 2 3 4 5 6 7

No. of relevés 5 6 8 30 16 17 11

Mercurialis annua 6 20r 67+ 13+ . . . .

Hedypnois rhaga-dioloides

6 . 33+ . . . . 36+

Lotus halophilus 6 20r . . . . . 452

Trifolium tomentosum 6 . . 13r . . 24+ 9+

Torilis nodosa 5 60+ . 252 . . . .

Silene colorata 5 . . 13r . 6+ . 27+

Galium spurium 5 20+ . 251 7+ . . .

Urtica urens 5 . . . 3+ . 24+ .

Asparagus aphyllus subsp. orientalis

4 . 501 13+ . . . .

Filago pyramidata 4 20r . . . . 6r 18+

Catapodium rigidum 4 40+ . 13+ . . . 9+

Prasium majus 4 20r . 25+ 31 . . .

Campanula erinus 4 . . . 7+ . . 18+

Phagnalon graecum 4 . . . 7r 6+ 6+ .

Galium aparine 3 . 33+ . 3r . . .

Anthemis werneri subsp. werneri

3 . . 131 . . . 18+

Galium murale 3 . . . . . 61 182

Myosotis incrassata 3 . . . 10+ . . .

Briza maxima 3 . . 13+ 3+ 6+ . .

Fig. 2. Atriplex halimus–Lycium schweinfurthii community on the southern coast of PK. Intricate, halo-nitrophilous scrub on coarse lava blocks aged centuries, salt-sprayed (3 May 1984, photo T. Raus).

8 P. Dimopoulos et al.

Comm. 7: Lupinus angustifolius–Tolpis barbata (Helianthemetea guttati)

Distribution: PK (11 relevés, 4 sampled in 2000 and 7 in 1984).

Diagnostic species and structure: Assemblage of an-nuals of which, e.g., Andryala integrifolia, Crassula alata, Hypochaeris glabra, Psilurus incurvus, Tolpis barbata and a dozen others (see Tab. 1) show differ-ential focus of occurrence, and Corynephorus articu-latus, Filago gallica, Trifolium campestre and Vulpia ciliata and others are more frequent than Lupinus angustifolius.

Ecology: Young fi ne ashes; fi rst stage of colonisation before perennials enter the competition.

Environmental factors shaping vegetation patterns

The RDA of the entire data set (Fig. 4A) on the basis of seven explanatory variables showed that the main gradient (Axis 1) in the data set is structured by the age of the lava substrate. Relief and slope have high correlation between each other as well as with the RDA Axis 2. Altitude and distance from the shore are correlated to both of the RDA axes, while heat load and annual radiation were found having non-signifi cant unique contribution to the explanation of species data variance. Additionally, the latter vari-ables explain a small proportion of the variance even as stand-alone variables (Table 3). The total variance explained in this analysis is 23.7 % (Table 2).

The RDA on the data set of the Communities 4 through 7 (Fig. 4B) identifi ed four variables (lava age by present, altitude, relief, and slope) as having sig-nifi cant contributions to the explanation of the spe-cies data variance (Table 3). The age of lava substrate is correlated mainly to the RDA Axis 1, and the alti-tude, relief and slope are correlated to both axes. The total variance explained is 23.5 % (Table 2). In both RDA diagrams constructed using seven explanatory variables, the discrimination as well as the disper-sion of the groups of relevés (Communities) is fairly unclear because overlaps between vegetation units occur. Besides, the total variance explained in these RDAs is low (Table 2).

The use of the EIVs for light, temperature, mois-ture, reaction, nitrogen, and salinity improved sig-nifi cantly the interpretability and explanatory power

Fig. 3. Lupinus angustifolius–Helichrysum italicum community in the interior of NK. Non-halophytic, permanent matrix of the semi-shrub Helichrysum italicum on young (34 yr) ashes growing spaced because of underground root competition, accompanying thero-phytes seasonally withered. (7 May 1984, photo T. Raus).

Table. 2. Eigenvalues and total variance explained (TVE) by the signifi cant variables in the RDA of all relevés and of those of the Communities 4 through 7. Results of the analyses involving seven and 13 explanatory variables, respectively, are presented.

Relevés data setExpl. var. data set

EigenvaluesTVE (%)

Axis 1 Axis 2

All relevés 7 0.11 0.06 23.7Communities 4–7 7 0.104 0.083 23.5All relevés 13 0.126 0.089 46.9Communities 4–7 13 0.165 0.113 50.1

Vegetation patterns and primary succession on sea-born volcanic islands (Santorini archipelago, Aegean Sea, Greece) 9

Fig. 4. Biplots of relevés and explanatory variables scores along the fi rst two RDA axes. The upper two diagrams (A and B) were constructed on the basis of seven explanatory variables, while the lower two diagrams (C and D) were constructed on the basis of 13 variables. In the left-hand diagrams, the data set of all relevés was used, while in the right-hand ones the data comprising communities 4–7 were used. Only the signifi cant variables (Monte Carlo test with 999 permutations, p≤ 0.05) were used in the fi nal ordinations. For the explanation of the abbreviations of explanatory variables see Tables 3 and 4.

Table. 3. Results of forward selection in RDA using seven explanatory variables (LA-BP: lava age by present, DS: distance from the shore, Alt: altitude, Sl: slope, Re: relief, Rad: annual direct incident radiation, Heat: heat load). L1: variance explained by each variable singly, LA: additional variance that each variable explains at the time of its inclusion in the model, P: signifi cance level, F: F-ratio, Or: order of inclusion of variable into the model.

Variable L1 LA P F Or L1 LA P F Or

All relevés Communities 4 to 7

LA-BP 0.11 0.11 0.001 10.8 1 0.1 0.1 0.001 7.87 1DS 0.06 0.02 0.002 2.33 5 0.05 0.01 0.191 1.28 5Alt 0.05 0.04 0.001 4.1 3 0.05 0.05 0.001 4.82 3Sl 0.04 0.03 0.002 2.95 4 0.07 0.06 0.001 4.77 2Re 0.04 0.04 0.001 4.93 2 0.06 0.03 0.01 2.32 4Rad 0.02 0 0.869 0.63 7 0.03 0.01 0.449 0.97 6Heat 0.02 0.01 0.178 1.27 6 0.02 0.01 0.819 0.67 7

10 P. Dimopoulos et al.

of the RDA diagrams (Figs. 2C and 2D), and revealed more reliably the prevailing environmental gradients. The proportion of explained variance (Table 2) in-creased up to 47–50 %. The Axis 1 of the RDA of the entire data set (Fig. 4C) is negatively correlated to the age of lava substrate, and positively correlat-ed with the light intensity. The Axis 2 of the same analysis represented mainly nitrogen and reaction, and to a lesser extent also salinity and altitude. Three communities dominated by perennials found on PK are discriminated along Axis 1. Along the Axis 2 the halophytic from the non-halophytic communities of PK are discriminated as well as are the dense thick-ets of the Lupinus angustifolius–Hyparrhenia hirta comm. from the rest therophyte communities on both islands. In this analysis the variables such as dis-tance from the shore, heat load and annual radiation did not contribute signifi cantly to the explanation of species data variance (Table 4).

The discrimination of the therophyte communities (sub-set of the entire data set; Communities 4 to 7) is better illustrated in the RDA of Fig. 4D. In the lat-ter diagram the Axis 1 corresponds to the Axis 2 of the analysis depicted in Fig. 4C, while the Axis 2 dis-criminates the unique therophyte community of PK

from the corresponding ones of NK. In this analysis as non-signifi cant variables were found the same ones as in the previous analysis, joined by relief (Table 4).

Among all above-mentioned RDA analyses the age of lava substrate was selected as the fi rst or sec-ond variable. EIVs for light, nitrogen, reaction, and salinity explained higher proportions of variance than did the physiographic variables in the RDA analysis involving the entire data set. Nitrogen and reaction explained also the highest proportions of variance in the data set of the Communities 4 through 7 (Tables 3 and 4).

In order to test the plausibility of the RDA dia-grams, the indirect ordination method of PCA was applied using the same data sets. In these analyses the explanatory variables were used as passive ones and they just projected on the ordination space calculated singly on the basis of species occurrence in the relevés. The results of the PCA analyses (not presented here) were fairly similar to those of the RDA analyses where the 13 explanatory variables were used. The similarity relates to the spread of the relevés groups (Communities) on the ordination space as well as the correlation of the passive explanatory variables with the fi rst two PCA axes. Therefore, we conclude that

Table. 4. Results of forward selection in the RDA using 13 explanatory variables (those presented in Table 3 plus EIVs for light (L), temperature (T), moisture (M), reaction (R), nitrogen (N) and salinity (S). For other abbreviations see Table 3.

Variable L1 LA P F Or L1 LA P F Or

All relevés Communities 4 through 7

LA-BP 0.11 0.11 0.001 10.8 1 0.1 0.1 0.001 9.32 2DS 0.06 0 0.459 0.99 12 0.05 0.01 0.462 0.98 12Alt 0.05 0.02 0.001 2.8 9 0.05 0.03 0.001 3.6 7Sl 0.04 0.03 0.001 3.52 7 0.07 0.05 0.001 5.17 4Re 0.04 0.04 0.001 5.92 5 0.06 0.01 0.32 1.1 10Rad 0.02 0.01 0.605 0.87 13 0.03 0 0.382 1.05 13Heat 0.02 0.01 0.271 1.16 11 0.02 0.01 0.252 1.15 11L 0.1 0.07 0.001 9.07 3 0.05 0.01 0.019 1.94 9T 0.04 0.03 0.001 4.13 6 0.05 0.04 0.001 5.18 5M 0.04 0.05 0.001 5.75 4 0.07 0.07 0.001 6.19 3R 0.06 0.02 0.001 3.07 10 0.1 0.03 0.001 3.78 6N 0.08 0.08 0.001 9 2 0.13 0.13 0.001 11.1 1S 0.06 0.02 0.001 3.57 8 0.06 0.04 0.001 4.02 8

Table 5. Variance partitioning table showing the effects (proportion of explained variance) of the age of lava substrate and the other (physiographic and EIVs) variables, as well as the shared effects and the residual variation in the whole data set and in two subsets (Comms. 4–7: all therophyte communities; Comms. 4–6: communities of NK only).

Variance source All relevés Comms. 4–7 Comms. 4–6

Lava age by present 5.6 6.2 2.3Other variables 36.3 40.2 39.6Shared 5 3.6 9.6Residual 53.1 49.9 48.6

Vegetation patterns and primary succession on sea-born volcanic islands (Santorini archipelago, Aegean Sea, Greece) 11

the RDA performed on the basis of the 13 explana-tory variables revealed well the environmental gra-dients affecting the vegetation differentiation on the studied islands.

RDA showed that the age of lava substrate is the most important factor affecting the composition of the plant communities in the two islands. In order to explore better its role, a variance portioning was ap-plied by means of partial RDA. In these analyses only the variables having a unique signifi cant contribution to the explanation of variance were used. Variance partitioning was applied in three data sets: the set of all relevés, the therophyte communities (Communi-ties 4–7) and the communities of NK. The data set of the 13 explanatory variables was used. The results of variance partitioning are presented in Table 5. In all the three data sets, the variance explained by lava age by present alone as well as the shared variance is signifi cantly smaller than the variance attributed to the other variables. Although the sum of the shared variance and that attributed to lava age in the data set of NK plant communities becomes higher than the corresponding sum in the other two data sets, it is unexpectedly small taking into account the high vol-canic activity in this island over the last century.

Discussion

NK together with Anak Krakatau (Indonesia) are two of the world’s most prominent examples of ma-rine emergent islands, unique and typical examples of primary mediterranean and tropical succession, respectively (WHITTAKER et al. 1989, THORNTON & NEW 2007). Cases of natural colonization and com-munity assembly of terrestrial organisms on a macro-scale from a zero base-line are few, as such oppor-tunities are very rare. Interest in the colonization of emergent islands and the recolonization of devas-tated islands leading to the origin of biotic commu-nities is both intrinsic (e.g. BRATTSTROM 1963, BALL & GLUCKSMAN 1975, FRIDRIKSSON 1975, DIAMOND 1981, WHITTAKER et al. 1989, THORNTON 1996a,b, 2001) and theoretical (e.g. DIAMOND 1975, GILPIN & DIAMOND 1982, TAGAWA 1992, THORNTON et al. 1992, WHITTAKER 1992, BUSH & WHITTAKER 1993, WARD & THORNTON, 2000).

Floristic and structural differences between the vegetation of the two islands are striking. In contrast to NK, the existence on PK of structurally complex (several-layer) shrubby communities suggests a long-term (several hundreds of years) volcanic quiescent period. That mere age of surface is the basic prerequi-site for the stable development of the vegetation type dominated by Atriplex halimus (Chenopodiaceae) and Lycium schweinfurthii (Solanaceae) can be deduced from the observation that the former species was able to start colonization of NK (RAUS 1986: 390) not ear-lier than only shortly before 1984, i.e. roughly 35–40 years after the last volcanic event on NK (in 1950) on

a surface originating from a lava fl ow in 1866–1870. The latter species has not yet reached NK.

Lava age was found (by the RDA) as the most im-portant “ecological” structuring factor. The woody vegetation on PK is clearly linked to the age of the lava surface of PK which is offered to plant coloniza-tion for many centuries as compared to NK, which has been forming since 1570 onwards. Atriplex hal-imus tries to enter the island of NK, occupying its typical niche of salt-sprayed rocks near the seawater line, whereas Pistacia lentiscus does the same on top of NK, as far away as possible from salt-spray infl u-ence (see RAUS 1986 for details). Both woody taxa have diffi culties to establish themselves on NK; the lack on NK of more or less closed shrubby vegeta-tion of the Atriplex halimus–Lycium schweinfurthii and Pistacia lentiscus community types is obviously caused by the generally younger surface of NK.

The evergreen sclerophyllous shrub of the Pistacia lentiscus community represents the fi nal stage of the climatically-dependent vegetation succession, where-as the halophytic Atriplex halimus–Lycium schwein-furthii comm. represents the presumably fi nal stage of the supralittoral vegetation succession according to the prevailing environmental and fl oristic condi-tions on PK.

On NK, vegetation has reached different stages of succession in different parts of the island. The most evolved and closed vegetation type on NK at present, is the steppe-like Lupinus angustifolius–Hyparrhenia hirta comm. on young, shallow fi ne ashes, whereas on young coarse ashes rich in lava rocks, the Lupinus angustifolius–Umbilicus rupestris comm. is the fi nal stage. As far as the establishment of shrubby vegeta-tion in NK is concerned, we consider it as a matter of determinism (see THORNTON & NEW 2007, pp. 10–12) caused by the harsh environment of the arid mediterranean climate and the very immature soil on this island, rather than a matter of chance (stochas-ticity) related to the possibilities of shrub species to reach NK. However, in the variance partitioning the total variance attributed to lava age is far smaller than that attributed to the rest explanatory variables. This fi nding is also refl ected in the vegetation cover of the island of NK, where 300-years old lavas are very sparsely vegetated, whereas 50-years old ashes are densely vegetated by open-canopy herbaceous plant communities, even comprising single sub-shrub spe-cies, such as Helichrysum italicum. We hypothesise that in this last case, the driving force might not be the age of land surface, but texture of substrate ame-liorating the water holding capacity, hence allowing faster colonization by herbs and formation of more persistent plant cover.

Furthermore, we suggest that the chemical prop-erties of volcanic ash (especially soil acidity and nu-trient content) play an important role. According to TSUYUZAKI & DEL MORAL (1995) the development of woody vegetation on volcanic substrates is possible only under conditions of adequate nutrient sources. In the case of Vesuvius (active volcano towering over

12 P. Dimopoulos et al.

Napoli, central Italy) the plant colonization of tephra was much faster than that of hard-surface lava fl ows (MAZZOLENI & RICCIARDI 1993).

Taking into account that the fl ora of the two is-lands is comprised mainly of short-lived grasses and herbs as well as small proportion of dwarf shrubs and perennial herbs, the species pool for the re-vegetation of areas covered by ashes is much bigger than that of lava fl ows (see THORNTON & NEW 2007, p. 44). Although it is logical to expect that the age of sub-strate surface would play an important role in assem-bly of plant communities, the role of dispersal is ap-parently signifi cantly especially on the fi ne-grained volcanic substrates. Furthermore, one should argue that the high variance explained by the selected EIVs is a sequence of the fact that these are calculated on the basis of the fl oristic composition of relevés. The use of EIVs in direct ordination analyses is consid-ered to imply an element of circularity (DIEKMANN 2003). However, the similarity between the results of RDA and that of the indirect ordination method of PCA indicates that a circularity issue does not affect signifi cantly the results of RDA.

The Lupinus angustifolius–Hyparrhenia hirta comm. was known to exist already in 1984 and later, when it was documented from the fi ne-pumice and ash surfaces produced by the local volcanic eruptions of 1939–1941 and 1950 (RAUS 1986: 390). At that time (in 1984) this dense, grassy vegetation type was at most 43 to 34 years for it could not be older than the settled substrate. When two of us (PD & LM) again visited NK in 2000, the community of 1984 had de-veloped from a closed Lupinus–Hyparrhenia facies towards another stage of succession which did not exhibit the predominance of the two species as ob-served 16 years ago. This may be a matter of tempo-ral fl uctuations of soil resources, especially those of limiting availability. This community is clearly linked (according to our RDA analyses) to soil reaction and nitrogen availability. Contents of organic matter and nitrogen status of soil are among the most important limiting factors in the primary succession of volcanic substrates (THORNTON & NEW 2007, pp. 45–46).

The Lupinus angustifolius–Tolpis barbata comm. of PK is ecologically similar to the communities found in NK, but it houses many differential taxa (see Tab. 1) which occur exclusively on PK (at the scale of our data). This is, therefore, by far the species-richest community found on both islands. The latter fact is probably a refl ection of different surface age of both islands, since a bulk of species had not yet immigrat-ed to and established on the much younger NK. This may imply some stochasticity, as this community does not differ signifi cantly in its ecology from the therophyte communities of NK.

We suggest that the formation of local pioneer veg-etation on the studied volcanic islands is not solely controlled by the age of the substrate (“maturation” of the communities through saturation of their local species pools), but it is also affected by factors such as the distribution of ashes after the recent volcanic

eruptions, and it is rather a expression of control of chemical and physical (e.g. nutrient content) proper-ties of the soils formed by the deposition of ashes.

The arid environment of the studied islands, in contrast to Anak Krakatau where woody communi-ties were established in less than 30 years (WHITTAKER et al. 1989), seems to impose barriers to a fast devel-opment of vegetation cover and the establishment of woody communities in NK (after ca. 439 years). On the other hand, certain therophyte species that exist in the Lupinus angustifolius-Tolpis barbata comm. found on the much older PK, seems not to have suc-ceeded to arrive and establish into similar habitats of the Lupinus angustifolius communities (communities 4–7) in NK.

The island of NK and to a lesser degree PK, con-stitute a classic case study for both Mediterranean vegetation succession patterns and island biogeog-raphy, as is the respective case of tropical succes-sion in the Krakatau islands (WHITTAKER et al. 1989, THORNTON & NEW 2007). The colonization of Anak Krakatau since its emergence from Krakatau’s sub-marine caldera in 1930 has been studied over many years (THORNTON 1996a, b, 2000) and the early colo-nization of Motmot in Lake Wisdom, Long Island has been monitored since 1968 (BALL & GLUCKSMAN 1975, OSBORNE & MURPHY 1989).

In the islands of NK and PK, patterns of succes-sion will be studied through direct long term moni-toring using (a) permanent plots (installed in 1984 and 2000) that will be revisited in 2011, 100 years after the fi rst complete plant census on NK by VIERHAPPER (1914, 1919), (b) a sequence of closely located sites (of PK and NK) differing in age/ time after disturbance and substrate micro-topography, that alter local suc-cession rates by affecting environmental stress, and (c) detailed measurements of physical and chemical properties of tephra and ash deposits.

Acknowledgements: The Hellenic Ministry of the Environ-ment, Regional Planning and Public Works (YPEXODE), is gratefully acknowledged for funding PD and LM while survey-ing the vegetation of Santorini in 2000, in the framework of the project on habitat types mapping of the Natura 2000 sites in Greece. PD received support from the SYNTHESYS Project “Plant species colonization, dynamics and distribution patterns on sea-born volcanic islands in the Mediterranean (Aegean Sea, Greece)” (http://www.synthesys.info/) which is fi nanced by European Community Research Infrastructure Action un-der the FP6 “Structuring the European Research Area” Pro-gramme.

References

Albanakis, K. & Styllas, M. (2004): Variations of sea level and sea surface temperature in the Caldera of Santorini Island, southern Aegean Sea. – Bull. Geol. Soc. Greece 36: 1077–1083.

Ball, E.E. & Glucksman, J. (1975): Biological colonization of Motmot, a recently created tropical island. – Proc. R. Soc. Lond., Ser. B: Biol. Sci. 190: 421–442.

Vegetation patterns and primary succession on sea-born volcanic islands (Santorini archipelago, Aegean Sea, Greece) 13

Barkman, J.J., Doing, H. & Segal, S. (1964): Kritische Be-merkungen und Vorschläge zur quantitativen Vegetations-analyse. – Acta Bot. Neerl. 13: 394–419.

Böhling, N., Greuter, W. & Raus, T. (2002): Zeigerwerte der Gefäßpfl anzen der Südägäis (Griechenland). – Braun-Blan-quetia 32: 1–108.

Bond, A. & Sparks, R.S.J. (1976): The Minoan eruption of San-torini, Greece. – J. Geol. Soc. 132: 1–16.

Brattstrom, B.H. (1963): Barcena Volcano, 1952: its effect on the fauna and fl ora of San Benedicto Island, Mexico. – In: Gres-sitt, J.L. (ed.): Pacifi c Basin Biogeography. pp. 499–524. – Bishop Museum Press, Honolulu.

Braun-Blanquet, J. (1964): Pfl anzensoziologie. Grundzüge der Vegetationskunde. 3. Aufl . – Springer, Wien.

Bush, M.B. & Whittaker, R.J. (1993): Non-equilibration in is-land theory of Krakatau. J. Biogeogr. 20: 453–457.

Cammerloher, H. (1935): Ein Beitrag zur Flora von Kaimeni (Santorin). – Österr. Bot. Z. 84: 81–90.

Casale, R., Fytikas, M., Sigvaldasson, G. & Vougioukalakis, G. (eds.) (1998): The European laboratory volcanoes. Proceed-ings of the Second Workshop Santorini, Greece, 2 to 4 May 1996. – Offi cial Publications of the European Communities, Luxembourg.

Diamond, J.M. (1975): Assembly of species communities. – In: Cody, M.L. & Diamond, J.M. (eds.): Ecology and evolution of communities. pp. 342–444. – Harvard University Press, Cambridge, MA.

Diamond, J.M. (1981): Reconstitution of bird community struc-ture on Long Island, New Guinea, after a volcanic explosion. – Nat. Geogr. Soc. Res. Reports 1972: 191–204.

Diapoulis, C. (1971): The development of the fl ora of the volca-nic islands Palaia Kammeni and Nea Kammeni. – In: Kalog-eropoulou, A. (ed.): Acta of the 1st International Congress on Volcano Thera 1969. pp. 238–247. – Archeol. Service Greece, Athens.

Diekmann, M. (2003): Species indicator values as an important tool in applied plant ecology – a review. – Basic Appl. Ecol. 4: 493–506.

Dimitriadis, I.M., Panagiotopoulos, D.G., Papazachos, C.B., Hatzidimitriou, P.M., Karagianni, E.E. & Kane, I. (2005): Recent seismic activity (1994–2002) of the Santorini volcano using data from local seismological network. – Develop. Vol-canol. 7: 185–203.

Downey, W.S. & Tarling, D.H. (1984): Archaeomagnetic dating of Santorini volcanic eruptions and fi red destruction levels of late Minoan civilization. – Nature 309: 519- 523.

Druitt, T.H. & Francaviglia, V. (1992): Caldera formation on Santorini and the physiography of the islands in the late Bronze Age. – Bull. Volcanol. 54: 484–493.

Druitt, T.H., Davies, M.S., Edwards, L., Sparks, R.S.J., Mellors, R.M., Pyle D.M., Lanphere, M. & Barreirio, B. (eds.) (1999): Santorini volcano. Volcanism and the Natural History of Santorini. – Geological Society of London, Special Memoir 19: 1–165.

Druitt, T.H., Mellors, R.A., Pyle, D.M. & Sparks, R.S.J. (1989): Explosive volcanism on Santorini, Greece. – Geol. Mag. 126: 95–126.

Fouqué, F. (1879): Santorin et ses éruptions. – Masson & Cie, Paris.

Fridriksson, S. (1975): Surtsey: evolution of life on a volcanic island. – John Wiley & Sons, New York.

Friedrich, W.L. (2004): Feuer im Meer. Der Santorin-Vulkan, seine Naturgeschichte und die Atlantis-Legende. Ed. 2. – Spektrum, Heidelberg.

Friedrich, W.L., Eriksen, U., Tauber, H., Heinemeier, J., Rud, N., Thomsen, M.S. & Burchardt, B. (1988): Existence of a

water-fi lled caldera prior to the Minoan eruption of Santo-rini, Greece. – Naturwissenschaften 75: 567–569.

Friedrich, W.L., Seidenkrantz, M.-S. & Nielsen, O.B. (2000): Santorini before the Minoan eruption – a reconstruction of the ring-island based on geological and archaeological evi-dence. – In: McGuire, W.J., Griffi ths, D.R., Hancock, P.L. & Stewart, I.S. (eds.): The archaeology of geological catastro-phes. pp. 71–80. – The Geological Society, London.

Friedrich, W.L., Kromer, B., Friedrich, M., Heinemeier, J., Pfe-iffer, T. & Talamo, S. (2006): Santorini eruption radiocarbon dated to 1627–1600 B.C. – Science 312: 548.

Fytikas, M., Kolios, N. & Vougioukalakis, G. (1990): Post-Minoan volcanic activity of the Santorini volcano. Volcanic hazard and risk, forecasting possibilities. – In: Hardy, D.A., Keller, J., Galanopoulos, V.P., Fleming, N.C. & Druitt, T.H. (eds.): Thera and the Aegean World III. 2. Earth sciences. pp. 183–198. – The Thera Foundation, London.

Georgalas, G.C. (1953): L’éruption du volcan de Santorin en 1950. – Bull. Volcanol., Sér. 2, 13: 39–55.

Gilpin, M.E. & Diamond, J.M. (1982): Factors contributing to nonrandomness in species co-occurrence on islands. – Oeco-logia 52: 75–84.

Greuter, W., Burdet, H. & Long, G. (eds.) (1984, 1986, 1989): Med-Checklist. A critical inventory of vascular plants of the circum-mediterranean countries. 1, 3, 4. – Conservatoire et Jardin botaniques, Genève.

Hansen, A. (1971): Flora der Inselgruppe Santorin. – Candollea 26: 109–163.

Hennekens, S.M. & Schaminée, J.H.J. (2001): TURBOVEG, a comprehensive data base management system for vegetation data. – J. Veg. Sci. 12: 589–591.

Keller, J. (1980): Did the Santorini eruption destroy the Minoan world? – Nature 287: 779.

Legendre, P. & Legendre, L. (1998): Numerical ecology. – El-sevier, Amsterdam.

Legendre, P. & Gallagher, E.D. (2001): Ecologically meaningful transformations for ordination of species data. – Oecologia 129: 271–280.

Makris, K. (2006): The evolution of vegetation on and around Santorini volcano through the ages. A critical review on Thomas Raus’ “Vascular plant colonization and vegetation development in sea-born volcanic islands in the Aegean”. – http://cultureparks.org/php/articles.php?lng=gr&pg=104 (accessed on 05.06.2006).

Mazzoleni, S. & Ricciardi, M. (1993): Primary succession on the cone of Vesuvius. –In: Miles, J. & Walton, D.W.H. (eds). Pri-mary succession on land. pp. 101–112. Blackwell Scientifi c Publications, Oxford.

McBirney, A.R. (1999): Ferdinand A. Fouqué, Santorini and its eruptions, translated and annotated. (Foundations of natural history). – Johns Hopkins University Press, Baltimore.

McBirney, A.R. (2000): Friedrich, W. L., Fire in the sea. The Santorini Volcano: Natural History and the Legend of At-lantis. Ed. 1, translated. – Cambridge: University Press, Cambridge.

McCune, B. & Grace J.B. (2002): Analysis of ecological com-munities. – MjM Software Design, Gleneden Beach, OR.

McCune, B. & Keon, D. (2002): Equations for potential an-nual direct incident radiation and heat load. – J. Veg. Sci. 13: 603–606.

McCune, B. & Mefford M.J. (1999) : PC-ORD – Multivariate analysis of ecological data, Version 5.0. – MjM Software, Gleneden Beach, OR.

Mucina, L. (1997): Conspectus of classes of European vegeta-tion. – Folia Geobot. Phytotax. 32: 117–172.

14 P. Dimopoulos et al.

Osborne, P.I. & Murphy, R. (1989): Botanical colonisation of Motmot Island, Lake Wisdom, Madang Province. – Science in New Guinea 15: 57–63.

Papadopoulos, G.A. & Orfanogiannaki, K. (2005): Long-term prediction of the next eruption in Thera volcano from con-ditional probability estimates. – Develop. Volcanol. 7: 211–216.

Papazachos, B.C. (1990): Long and short term prediction of the volcanic eruptions in Santorini. – In: Hardy, D.A., Keller, J., Galanopoulos, V.P., Fleming, N.C. & Druitt, T.H. (eds.): Thera and the Aegean World III. 2. Earth sciences. pp. 224–228. The Thera Foundation, London.

Pichler, H. & Kussmaul, S. (1980): Comments on the geologi-cal map of the Santorini Islands (with a coloured Geologi-cal Map 1:20 000 and two profi les). – In: Doumas, C. (ed.). Thera and the Aegean World II. pp. 413–427. The Thera Foundation, London.

Pyle, D.M. & Elliott, J.R. (2006): Quantitative morphology, recent evolution, and future activity of the Kameni Islands volcano, Santorini, Greece. – Geosphere 2: 253–268.

Podani, J. (1994): Multivariate data analysis in ecology and sys-tematics. – SPB Publishing, The Hague.

Raus, T. (1986): Floren- und Vegetationsdynamik auf der Vulk-aninsel Nea Kaimeni (Santorin-Archipel, Kykladen, Griech-enland). – Abh. Landesmus. Naturkd. Münster/Westf. 48: 373–394.

Raus, T. (1988): Vascular plant colonization and vegetation development on sea-born volcanic islands in the Aegean (Greece). – Vegetatio 77: 139–147.

Raus, T. (1991): Die Flora (Farne und Blütenpfl anzen) des San-torin-Archipels. – In: Schmalfuss, H. (ed.). Santorin – Leben auf Schutt und Asche. Ein naturkundlicher Reiseführer. pp. 109–124. – Verlag J. Margraf, Weikersheim.

Reck, H. (ed.) (1936): Santorin. Der Werdegang eines Insel-vulkans und sein Ausbruch 1925–1928. Ergebnisse einer Deutsch-Griechischen Arbeitgemeinschaft. 1–3. – Reimer, Berlin.

Richter, M. (1984): Vegetationsdynamik auf Stromboli. Zur Geoökologie trocken-mediterraner Standorte. – Aachener Geogr. Arb. 16: 41–110.

Schmalfuss, H., Steidel, C. & Schlegel, M. (1981): Die Fauna der Ägäis-Insel Santorin. Teil 1. – Stuttgarter Beitr. Naturk. Ser. A, 347: 1–14.

Sparks, R.S.J. (2000): The causes and consequences of eruptions of andesite volcanoes: opening remarks. – Phil. Trans. R. Soc. London 358: 1435–1440.

Tagawa, H. (1992): Primary succession and the effect of fi rst arrivals on subsequent development of forest types. – Geo-Journal 28: 175–183.

ter Braak, C.J.F. & Šmilauer, P. (2002): CANOCO Reference manual and CanoDraw for Windows User’s guide: Software for Canonical Community Ordination (version 4.5). – Mi-crocomputer Power, Ithaca, NY.

Thornton, I. (1996a): Krakatau: the destruction and reassembly of an Island Ecosystem. – Harvard Univ. Press, Cambridge.

Thornton, I. (1996b): The origins and development of island bio tas as illustrated by Krakatau. – In: Keast, A. & Miller, S.G. (eds.): The Origin and Evolution of Pacifi c Island Bio-tas, New Guinea to Eastern Polynesia: Patterns and Pro-cesses, pp. 67–90. -SPB Academic Publishing, Amsterdam.

Thornton, I. (2001): Colonization of an island volcano, Long Island, Papua New Guinea and of an emergent island, Mot-mot in its caldera lake. I. General introduction. – J. Biogeogr. 28: 1299–1310.

Thornton, I.W.B & New, T.R. (2007): Island colonization: The origin and development of island communities. – Cambridge University Press, New York.

Thornton, I.W.B., Ward, S.A., Zann, R.A. & New, T.R. (1992): Anak Krakatau: a colonization model within a colonization model? – Geojournal 28: 271–286.

Tichý, L. (2002): JUICE, software for vegetation classifi cation. – J. Veg. Sci. 13: 451–453.

Tichý, L. & Holt, J. (2006): JUICE: Program for management, analysis and classifi cation of ecological data. – Vegetation Science Group, Masaryk University Brno.

Tsiripidis, I., Bergmeier, E. Fotiadis, G. & Dimopoulos P. (2009): A new algorithm for the determination of differential taxa. – J. Veg. Sci. 20: 233–240.

Tsuyuzaki, S. & del Moral, R. (1995): Species attributes in early primary succession on volcanoes. – J. Veg. Sci. 6: 517–522

Tutin, T. G., Burges, N. A., J. R., Heywood, V. H., Moore, D. M., Valentine, D. H., Walters, S. M. & Webb, D. A. (eds.) (1968, 1972, 1976, 1980): Flora Europaea. 2, 3, 4, 5. – Univer-sity Press, Cambridge.

Tutin, T. G., Burges, N. A., Chater, A. O., Edmondson, J. R., Heywood, V. H., Moore, D. M., Valentine, D. H., Walters, S. M. & Webb, D. A. (eds.) (1993): Flora Europaea. 1. Ed. 2. – University Press, Cambridge.

Vierhapper, F. (1914, 1919): Beiträge zur Kenntnis der Flora Griechenlands. Bearbeitung der anlässlich der zweiten Wie-ner Universitätsreise im April 1911 in Griechenland gesam-melten Pfl anzen. A. Anthophyta und Pteridophyta. – Verh. Zool.-Bot. Ges. Wien 64: 239–270, 69: 102–312.

Ward, S.A. & Thornton I.W.B. (2000): Chance and determinism in the development of isolated communities. – Global Ecol. Biogeogr. Lett. 9: 7–18.

Washington, H.S. (1926): Santorini eruptions of 1925. – Geol. Soc. Amer. Bull. 37: 349–384.

Whittaker, R.J. (1992): Stochasticism and determinism in island ecology. – J. Biogeogr. 19: 587–591.

Whittaker, R.J., Bush, M.B. & Richards, K. (1989): Plant recolo-nization and vegetation succession on the Krakatau Islands, Indonesia. – Ecol. Monogr. 59: 59–123.

Zuur, A.F., Ieno, E.N. & Smith G.M. (2007): Analysing ecologi-cal data: Statistics for biology and health. – Springer, New York.

Addresses of the authors:Panayotis DIMOPOULOS*, Department of Environmental and Natural Resources Management, University of Ioannina, Seferi 2, 30100 Agrinio, Greece.Thomas RAUS, Botanischer Garten und Botanisches Museum Berlin-Dahlem, Freie Universität Berlin, Königin-Luise-Straße 6–8, 14191 Berlin, Germany.Ladislav MUCINA, Department of Environmental Biology, Curtin University of Technology, GPO Box U1987, Perth, WA 6845, Australia.Ioannis TSIRIPIDIS, Department of Botany, School of Biol-ogy, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece.*Corresponding author, e-mail: pdimopul@cc.uoi.gr