VERTICAL LANDSCAPE STRUCTURE OF THE SOUTHERN PART OF VIS ISLAND , CROATIA

The paper presents some basic features of vertical landscape structure of the southern part of Vis Island, Croatia. Its aim is the determination of geocomplex types with a certain degree of stability and resistance to external influences, and confirmation or rejection of hypothesis that with the application of appropriate methods, the spatial relation between geocomplex types as well as the identification of specific dominant/stable and vulnerable/labile geocomplex types can be precisely determined. The results should serve as the basis for estimation of current status and future trends in the development of geocomplex types as well as the environmental changes.


I. INTRODUCTION
In landscape element investigations, hierarchical approaches to classification are often used (Zonneveld, 1989).By measuring and analysing, based on pre-selected criteria, it is possible to identify homogeneous landscape eiements to a greater or lesser extent (depend- ing on the scale).The criteria are represented by dilferent quantitative or qualitative environmental variables (geological, geomorphological, climatological, pedological and vegetational leatures as well as the agricultural/urban/transportation features ofland use and the historical-geographical characteristics ofthe study environment), that will correspond to the demanding geoinformatic criteria in the further analysis (Forman and Godron, 1986;  Zonneveld, 1989; Culotta and Barbera,2010).
Karst systems al1 over the world are extremely fragile and susceptible to all kinds of external shocks that cause irreversible changes (Ford and Williams, 2007).Considering internal abiotic and biotic differences, karst areas of the Adriatic islands present a mosaic of different landscapes, which is particularly expressed in the area of Vis Island.
Investigated area includes the southern part of Vis Island (20.86 km') (Figure 1) and it has been selected because ofits exceptional bio/geodiversity ofthe natural environtnent.'uo\\\ Studied area can be divided in two morphostructurally distinctive parts: northern part dominated by poljes, and southern part which represents limestone mountainous area without significant recent agricultural production and almost unpopulated.
During the historical development and even today, geographical isolation of Vis Island has affected its socio-economic development.Demographic aging and depopulation, especially in the 20th century (Neja5mi6 and  Mi5eti6, 2006) influenced the changes in landscape ofthe island (intensifying the process ofvegetational succession and dry stonewalls degradation).
In this research, vertical landscape structure of landscape units (geocomplex types) was analysed.Landscape types have been determined on the basis of their abiotic (lithological and geomorphological features) and biotic elements (natural and cultivated vegetation), taking in consideration the human impact during the previous time (field cultures on cultivated land and in urbanised areas).
By using GIS tools, overlapping of three parameter layers related to abiotic and biotic features (lithology, slope inclination and vegetation types), 2556 basic units (geocomplexes)   have been obtained.By generalization according to the similarity principle of the component features, 132 geocomplex types have been determined (2556 individual geocomplexes classified into -l32 types).These types represent generalized homogeneous spatial units which remained basis to all further analyses.In the next stage ofinvestigation, vertical structure features have been determined by the relation between geocomponents (lithological structure, vegetation types and slope inclination) and analysed for each geocomplex type separately.
The final goal ofthis research is the confirmation or rejection ofthe hypothesis that by the application of the above mentioned procedures, it is possible to precisely determine the spatial relations between the geocomplex types as well as the existence of specific dominant/stable and vulnerable/labile geocomplex types.The results should serve as the basis for estimation of the current status and future trends of geocomplex types development as well as environmental change in general, in positive or negative direction, which can be applied in future planning and protection ofthe investigated area.

RESEARCH METHODS
For the analysis of the vertical structure of the landscape, three parameters were used: lithology, vegetation and slope inclination.Each parameter consists ofseveral classes (slope 5 classes, vegetation l0 and geology 5), and each class consists ofelements ofdifferent sizes, which are present at various locations within study area.
By the overlapping of the parameters, new synthetic elements (geocomplexes), contain- ing common attrlbutes of parameters, were obtained.Due to a large number of elements (2556 combinations within parameters) and in order to manipulate data easily, in neu'overlapped layer the following actions were done: . connecting ol elements with identical attributes (132 synthetic classes (geocompler types) were obtained), and ' reclassification (numeric value/code was added to the attributes within overlapped layer, e.g., to the 'Lower Cretaceous limestones/garrigue-macchial12"32"' geocompler type value/code 33 was added).
Since at the landscape level the vertical structure cannot be displayed spatially, horizontal structure (spatial distribution ofgeocomplex types) served as a basis for analysis.That enabled the comparison and determination of hierarchy of connectivity strength of complete landscape.Connectivity strength index was used to determine the correlation between pairs of vertical structure elements (e.g., lithologyvegetation, lithologyslope, slope vegetation) and served as an indicator ofstability/instability ofthe parameters relationship within geocomplexes.
Slope inclination parameter was derived from a digital elevation model (DEM), which was created using semi-automatic and manual vectorization of contour lines of topographic maps (scale 1 : 25,000).WinTopo (tool for raster vectorization; Winlopo, 20121'Taie et a1., 2011; Dharmaraj,2005) was used for vectorization and this process consisted ofseveral steps: (1) colours sampling, (2) noise removal, (3) skeletisation Zhang Suen algorithm was used, (4) detection ofedges, (5) connection ofvector lines and (6) converting contour lines to shapefile.Attribute values (altitude) were added to converted layer.To obtain continu- ous surface with a series of z-values, contour lines were interpolated using triangulation irregular network (TIN) method (Mitas and Mitasova,1999, Webster and Oliver,2001;   Isenburg et a1., 2006; Jordan, 2007).Vector terrain model was produced and converted to raster for easier calculation of slope, exposition and elevation.GIS software calculates the slope inclination by using 3 x 3-squares method and calculates the maximum rate of altitude change of the central cell to the neighbouring cells (Burrough and McDonnell, 1998).Spa- tial resolution of the raster model was determined by cartographical rule method (Hengl,   2006), and was 12.5 meters.
Vegetation was manually vectorized (based on analysis of ARKOD, 2012, orthophoto maps in scale I : 5,000).According to species composition, ten vegetation classes were determined.During process of vectorization special attention was paid to the topological relations between classes (in order to avoid problems when overlapping with other param- eters).

GEOLOGICAL FEATURES
Structurally, Vis Island represents an anticline.The anticline core is formed of clastic sediments with gypsum and anhydrite in association with pyroclasts, as well as spilites and the diabases of Upper Ladinian Upper Norian age, while the limbs are formed of carbonate sediments (limestones and dolomites) of Cretaceous age (Borovi6 et al., 1917).
According to the hydrogeological properties, Terzil Q004) distinguishes several basic groups ofrocks: . Neocomian dolomites with low permeability and porosity covering relatively narrow zone around waterproof clastics and magmatites of KomiZa Cove (the contact between these two units is tectonic) (Borovi6 et a1., 1977); . carbonate rocks with medium permeability and cracking-dissolution porosity, calcareous dolomites, limestones ofCenomanian-furonian age, limestones and dolomitic limestones of Berriasian age with marls and marly layers, and limestones of Aptian-Albian age which form the largest part of the area (Borovi6 et al., 1917 ).They are partly karstified and porous enough to allow relatively rapid infiltration of rainwater into the ground; carbonate rocks of high permeability and cracking-dissolution porositywhite lime- stones of Senonian age, partly rudist limestones of Turonian age and karstified limestones of Cenomanian-Turonian age.Retention of the water in these fractured and karstified rocks is very limited, depending on the location; Quaternary rocks and sediments ofalternating properties, interseed and cracking poros- ity aeolian sands, terra rossa, colluvium breccias and conglomerates.
Geological mapping of the investigated area was conducted by Terzi6 (2004), but the authors made GIS analysis and calculated total areas (surfaces) for each lithological unit.
Figtu"e 3: Geological map o/ the southern part ofVis Island (according to Terzic, 2001)   Slika 3: Geoloika karta.juZnegadelo otoka Vis (po Terzit, 2004)   [*] Breccia and conglomeraies f_l terra rossa with rock fragments [] Catcitic doiomites of upper Cretaceous age illll sanas ffi Limesiones of upper Cretaceous age ----Faults -Major faults 4. GEOMORPHOLOGICAL FEATURES Southern part of Vis island is dominated by altitudes up to 200 meters above sea 1eve1 (Figure 4).Limestone mountain of Hurr (587 m) dominates the northwestern part of the study area, while the negative relief forms (poljes) are formed on dolomites, mostly along the taults.The slope ofsouthern coast is dissected by gullies and dry valleys.
Slope inclinations mostly reflect morphostructural relief features of the southern part ol \/is Island.Five categories have been allocated.The largest part, 48.8% (10.191 km'), belongs to the category of 12 to 32 degrees, followed by inclinations ofthe category 5 to representation of 13.4o/o Q.785 km2).Slopes with inclinations higher than 20 degrees cover 9.1% (1.9 km') of the observed area, while the smallest part belongs to the inclinations higher than 32 degrees with l.4o/o (0.286 km'z).Denudational processes become very active on the slopes steeper than 12 degrees (including the activation or increase oferosional processes, attrition and mass wasting).Denudational processes are characteristic for the slopes over 12 degrees and are more intensive on the slopes more exposed to the sunshine during the daylyear (S, SE, SW I.z il z-s * s-r: ffi re-se I'ga expositions) because of the modilication of solar radiation inlluences.That means increased temperature amplitudes which cause stronger mechanical r,veathering of the rock mass and soil drought (especially in summer) rvhich have negative inf-luence on vegetation cor,er.

VEGETATION FEATURES
Abiotic features of ecosystem along with the anthropogenic influence in the past and nowadays, have influenced the composition and distribution of characteristic vegetation species and associations of the southern part of Vis Island.The largest part of the area (18.3%) is covered by homogeneous or mixed areals of associations of evergreen fbrests, macchia, garrigue arrd bare rock with sparse grass vegetation.Very significant terraced agricultural areas in the past are nowadays in different overgrou,th stages, along u,ith the natural vegetation cover and fbrm a mosaic structure in the largest part ofthe investigated area.It mainly refbrs to Quercus i/e-x fbrests, which have been, due to the long history of human presence on Vis Island, significantly change<j (reclamation, fires and other negative factors).
Aleppo pine forests with larger or smaller proportion of holm oak (Qtterco ilicis-Pin- etum halepensr; Loisel, 1971) within the area with xerothermal climate, occupy microcii- matically more humid habitats.Forest and holm oak macchia with myrtle (Myrto-Quercettrnt ilicis; Trinajsti6, 1985) is the most thermophilic association, developed in areas u,ith favourable ecological conditions, which primarily relate to the temperature range during winter (average minimum of the coldest month between 6 and 8 'C) and sufficient Eeci(tation (average around 1000 mm per year, with a maximum in the colder part of the year).ln higher areas, where the conditions are colder and more humid, forests oiholoak and hop hornbeam (Ostryo-Quercetum ilicis) are widespread.
During the historical-geographical development, degradation ofautochthonous forests occurred because ofthe excessive and irrational logging and grazing, often in areas where, due to drought and temperature, significant soil retention and vegetation regrowth was not possible.Destructive fires (spontaneous or caused intentionally to obtain new areas for cultivation of crops; Gams,i991) need to be added to above mentioned factors.The degra- dation degree depended on the terrain morphology, soil characteristics and accessibiiity.The most degraded areas were located around the settlements on the upper parts of slopes surrounding poljes, where the original forests have almost completely disappeared due to Iogging and overexploitation.Nowadays, forest degradation is reduced to a minimum.
Macchia formed by forest degradation remained preserved (dense and almost completely impassable) in more isolated areas, often alternating with the holm oak or Aleppo pine forests.
Garrigue (further degradation stage) was dominantly created under influence of anthro- pogenic impact (gtazing,logging and fires) or by natural progression of the former rocky pastures, in the areas with shallow soil and exposed to strong insolation and drought in sum- mer.On the largest part of the area, garrigue is in association with other vegetation types, e.g., on abandoned agricultural land (mostly former vineyards) and is in association with the further degradation stage, which includes eumediterranean, stenomediterranean and rocky pastures.Today, large garrigue areas are mixed with holm oak macchia, or overgrown by Aleppo pine forests.In some areas, garrigue remains in the same degradation stage due to the unfavourable abiotic habitat conditions (e.g., very shallow and stony soil).
Rocky and bare ground areas prevail on the southern coastal slopes exposed to the wind influence (sirocco).On such surfaces, scarce shrub and grass vegetation are mostly present.Shrub vegetation occurs sporadically, mainly in less exposed areas (gullies), where the small part of soil remained, so these areas occasionally look like garrigue.Rarely and individually, Aleppo pine trees or smaller groups of other tree species occur.
Still active agricultural areas are mostly in poljes and valleys near the settlements.The most extensive are in the Dradevo and Plisko polje and other smaller poljes, representing the mosaics of different crops, mostly vineyards.Abandoned agricultural lands are present almost everywhere.These areas, once under vineyards and orchards, are found mainly on terraces built on steeper slopes in whole investigated area.
By analysing satellite images of the study area (ARKOD, 2012), different vegetation areas (natural or anthropologically modified) have been allocated.Ten types ofvegetation cover have been established (including the cultivated agricultural and agricultural/urbanized land categories) which occur homogeneous or in different interrelated combinations (Figure 7); . forest (abbr.F); ' combination of forest and macchia (as in Figure 7 and elsewhere in text higher propor- tion of forest, abbr.F/M); combination of macchia and forest (higher proportion of macchia, abbr.M/F); macchia (abbr.M); combination of macchia and garrigue (higher proportion of macchia, abbr.M/G); combination of garrigue and macchia (higher proportion of garrigue, abbr.G/M); garrigue (abbr.G); bare rock and sparse grass vegetation (abbr.BR/SGV); cultivated agricultural land in poljes (abbr.AA); cultivated areas on the slopes and urbanized land (abbr.AUL).Gatrigue / macchra (GiM) f, Forest/macchia(F/M) catrigue (G) iillllllli Macchia / forest (M/F) ffi eare rock withsparsegrass vegetation (BRtsGV) m Macchia (L'l) , I I Agriculture area (AA) $ Macchia I garigue (MlG) Agricultural and urbanized land (AUL)   6. VERTICAL LANDSCAPE STRUCTURE Vertical landscape structure analysis represents a method by which it is possible to de- termine relations between geocomponents contained in each geocomplex type (Kurnatowska, 1998) and can be expressed as connectivity strength index (W).This index is based on the relation between real surfaces with specific combinations of geocomponent features and theoretical, maximum surface on which these combinations can exist.Geocomponents analysed in this research include lithological characteristics, slope inclination and vegeta- tion characteristics.
Determination of interaction features between geocomponents within each geocomplex type (vertical structure) is essential for understanding of dominance and stability within geoecosystem.Some combinations of geocomponents appear more frequently and cover larger areas (indicating a greater stability and resistance ofgeoecosystem to external influences), while some appear rarely or not appear at all (low stability degree and high sensi- tir itr': Richling, 1992; Kurnatowska, 1998).Apart from fiequencies, there are some other iactors that can be used as indicators of stability/instability.These factors are mostly of There is no relationship between the following geocomponent pairs: F (forest)-S (Qua- ternary sands), FM (forest and macchia)-S, MF (macchia and forest)-S, M (macchia)-S, GM (garrigue and macchia)-S, G (garrigue)-S, BR/SGV (bare rock and sparse rock vege- tation)-S, AA (agricultural areas in poljes)-S, GTTRRF (terra rossa with rock fragments), BR/SGV-TRRF, MF-BC (breccias and conglomerates), G-BC, BR/SGV-BC.AA-BC.AUL (cultivated areas on slopes and urbanized land)-BC and M/BC.Very 1ow relationshrp (W: 0.001-0.2) characterises following geocomponent pairs: FM-TRRF, AA-LUC, MF-TRRF.F-TRRF, F-BC, GM]TRRF, M-TRRF, MG-TRRF, AA-CD, GM-CD, MF-CD ANd AUL-TRRF (Figure 8).As we mentioned above, some combinations of geocomponents appear rarely or not appear at all, indicatrng basic inadequacy between certain kinds ofvegetation cover and lithology, and thus, low stability degree and high sensitivity.
6.2.The relationship between vegetation cover and sloPe inclination Some studies (e.g., Kurnatowska, 1998; Kozlowska er a1.,2006') have shown clear links between vegetation cover and the type of morphodynamic surface (expressed through variations in the geomorphology of slopes).The majority of plant communities occur on a certain relief type and provide the natural boundaries for vegetation landscape units.Also, the different types olmorphodynamic units are characterized by particular t1,pes of vegetation (Kozlowska et a1., 2006).
Based on above mentioned facts, we can conclude that certain types ofvegetation cover appear more frequently on siopes of certain inclination.Example are crops that always occur on slopes with inclinations <2" or 2-5o.This is logical, because due to denudation processes, thicker soil layers could only be developed on slopes with 1ow inclination.On steeper slopes (12-32o), due to more pronounced denudation processes, soil layer is thinner and consequently, only combinations of macchia and garrigue or forest and macchia appear.

CONCLUSION
A detailed insight into the interrelation between the vertical landscape structure and the vertical connection of geocomponents of geocomplex types was provided by the comparative analysis and synthesis ofvertical landscape structure parameters ofthe southern part ofVis Island.This approach allowed determination of the stability degree ol each geocompler type and the determination of the most stable and most dominant types, as well as the most unstable and the most sensiti\e ones.
Due to the relatively unfavourable physical-geographical conditions, these areas are uninhabited and hurnan impact on the landscape is minimal.Expressed adaptive abilities (n.rorpho-anatomic, subcell and physiological-biochemical adaptations to external condi- tions) ofthe existing vegetation and the absence olnegative anthropogenic impact are the main reasons lbr preserving landscape balance.Because ofthe physical-geographical con- ditions, changes in ecosystems in terms of progressive succession of garrigue to macchia could not be expected to a significant extent.
The second geocomplex type according to the size (No.112) includes higher parts of Dradevo and Plisko polje, which have n.rainly anthropogenic soils with vineyards.There has been a long-term human impact on the natural landscape transformation, which partly reflects today's landscape appearance as well.Agriculture (especially viticultr-rre) prevails and due to intensive agricultural use in the past these areas have not been urbanized.Based on the vertical structure indicators, as well as field researches. it can be concluded that there is a balance between geocomponents and anthropogenic impact.This means that the land use in the past respected the natural conditions, while nowadays large parts ofthe area are abandoned in terms ofagricultural usage.It is left to natural process ofrenewal and succes- sion, which can further increase the stability of this geocomplex type.However, sometimes.it is not absolutely clear that the abandonment of agriculture increases the stability of s),srem because natural factors (especially soil type) have a significant role in this process.This can be seen in areas where intensive soil erosion is present (e.g., soil erosion in flysch oflstria is still very strong after many decades; Zorn and Petan, 2008).Height differences are small and the entire area is located at around 100 m above sea level.
Anthropogenic soils prevail and because of the intensive agricultural use in the past and recent reduction ofanthropogenic pressure, the area ofthis type is very similar to the type I l2 by its characteristics.
For the most stable geocomplex types, a strong connection between vertical geocompo- nents (lithology, vegetation and slope inclination) is characteristic while indicating a high degree of internal cohesion (Kurnatowska, 1998;Kozlowska et al., 2006).This is of grear importance because the internal cohesion directly affects the ecosystem resistance to ne_sa- tive external influences.
The allocation of geocomplex types with low stability and high sensitivity degree is very important, because it enables more efficient current and future protection ol their geoecosystems.When the changes occur within these geocomplex types (due to the ne_satir e impact ofnatural and anthropogenic environmental factors), they are often irrer ersible and.
if it comes to regeneration, a long period ofrecurrence time (in most cases, the s)'stem cannLr: b_ 5a nJa Lozii, Ante Siljeg, Kristina Krklec, Silvija 5ilieg / Dela 37 o 2012 r 65-90 return to previous state) is usually required.The reasons to that are mostly signilicant losses olpedological and/or vegetation cover.
By application of the comparative analysis and synthesis parameters of the vertical landscape structure (veryweakvertical correlation ofone ormore geocomponentpairs), twenty most unstable and most endangered geocomplex types (Figure 14 with a thin layer of terra rossa or lithosols.All three areas are located on the slopes of a gully which leads to cove, with a low altitude (50 m) and a large range of slope inclination (5 32).They are exposed to the south, southwest and southeast, and therefore exposed to a strong negative sea and wind (sirocco) influence.Related to this, in these areas, continuous sediment erosion and denudation processes are present, along with the strong anthropogenic influence expressed through apartment building construction and infrastructure (roads).
types, we can expect that over time the existing vegetation will change and adjust tt'r efl'r t- ronmental conditions.As a result of these processes, it could come to the transfbrmation trt hese geocomplex types into more stable ones, or, in the u'orst scenario, to a complete disappearanceofvegetationcoverandstrongactivationoferosionanddenudationprocesses, \ccording to some authors (Reice, 1994; Marston, 2010), consequences ofthese processes are not necessarily negative.Namely, the active erosion and denudation processes on the slopes in certain circumstances, could create positive conditions for recolonisation of species and i ncrease landscape heterogeneity.
Landscape structure is exposed to continuous change due to various activities related to spatial planning and management.The landscape represents an interface between natu-ra1 and social processes in the environment, while planning and decision-making related Figttre l5: Exctmples of the smallest and most sensitive geocomplex types (No. near Mala Tt'avna cove (ARKOD, 2012)   Sliko 15: Primer najmanjiih in najobifiljivejlih tipov geokompleksov (it.96, zaliva Mala Trarna (ARKOD, 2012)   96, 1l  (Turner, 1989).This is particularly important fbr karst areas, which are very sensitive to external influences due to their specific abiotic and biotic characteristics.
Estimation of the negative anthropogenic impact on these areas is a difficult task.For that reason, there is a need for development of multidisciplinary methods and techniques, by rvhich the changes in the environment of karst areas could be more efficiently determined (De Waele, 2009).
Natural balance disturbances have strong impact on geoecosystems and landscape as a ri'hole; therefore many ecological processes depend on the current dynamics of abiotic and biotic elements, considering the anthropogenic influence as well.The nature of these rela- tions is essential and often the result ofperiodic or episodic changes oflandscape features, that consequently affects bio/geodiversity.Environmental management strategies should take into account these changes in the dynamics of landscape elements.ln the study area, a better understanding ofcharacteristics ofverticai landscape struc- liirc sl.rould allow more efficient detection of changes related to tl.re natural landscape dy- :l.llttic and anthropogenically caused disturbances, which can lead to transition of geocom- : :l trpes ti'om the natural balance state to an imbalance state and to the increase of their '.,-rer:rbiLit1, as rvell. )- The primary task of this research was the establishment of appropriate methodology with an objective of exact analysis and synthesis of vertical landscape structure, which should improve the understanding of geoecological context of the current landscape state as well as the predictions of future development trends.Determination of the dominant, stable and resistant geocomplex types, and moreover the unstable, sensitive and non-resistant ones, could be a useful reference during the planning process and decision-making related to planning purposes and sustainable land-use in the southern part olVis Island.This primarily refers to the preventing ofexcessive exploitation ofnatural resources (vegetation cover devastation, quarrying and mining activity), inappropriate planning in urbanized zones, industriai and transport infrastructure, inadequate agricultural use and environmental pollution.
(Translated by Mtjo Zupit) limestones), GM-LUC (garrigue and macchia/Upper cretaceous limestones), BR/SGV- LUC (bare rock and sparse grass vegetation/Upper Cretaceous limestones), G-LUC (garrigue/Upper Cretaceous limestones), F-CD (forest/Upper Cretaceous calcitic dolomites) and AUL-CD (cultivated areas on the slopes and urbanized land/Upper Cretaceous calcitic doiomites; Table 1;Figure 8).Due to the fact that combinations of these geocomponents appear frequently and cover larger areas, we can assume that they indicate a greater stability and resistance of geoecosystem to external influences.This fact has been confirmed by some other investigations on other locations in Europe (Kurnatowska, 1998; Kozlowska et   a1., 2006).

Figure ll :
Figure ll: Most stable antl dontinont geocotnplex 4t1cs in are explained in lext)

Figure lj :
Figure lj: Dominant geocomplex type No. l12 located inwestern part of Draievo polje (ARKOD, 20121 Slika 13.Dominantni tip geokompleksa it.l2 v zahodnem delu Draievskega polja (ARKOD, 20121 ) have been determined.Several examples analysed rn detail during the lield research are shown in figures 15 and 16.Parts of the geocomplex types No. 96, 1 I and 41 (shown in Figure I 5) are situated near Mala Travna cove on the south coast.The bedrock is Upper Cretaceous limestone, covered