Catena 113 (2014) 281–291

Contents lists available at ScienceDirect

Catena

j ourna l homepage: www.e lsev ie r .com/ locate /catena

The characteristics of cultivated soils developed from coastal paleosand(Korcula Island, Croatia)

Romic M. a,⁎, Bragato G. b, Zovko M. a, Romic D. a, Mosetti D. b, Galovic L. c, Bakic H. a

a University of Zagreb Faculty of Agriculture, Dept. of Amelioration, Svetosimunska 25, 10000 Zagreb, Croatiab C.R.A. — Centro di Ricerca per lo Studio delle Relazioni tra Pianta e Suolo, Via Trieste 23, 34170 Gorizia, Italyc HGI-CGS — Croatian Geological Survay, Sachsova 2, P.O. Box 268, 10000 Zagreb, Croatia

⁎ Corresponding author. Tel.: +385 12394014; fax: +3E-mail address: mromic@agr.hr (R. M.).

0341-8162/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.catena.2013.08.009

a b s t r a c t

a r t i c l e i n f o

Article history:Received 17 May 2012Received in revised form 23 July 2013Accepted 12 August 2013

Keywords:Soil geochemistryGeostatisticsModal analysisSoil fertilityAdriatic coastSoil formation

The problem of maintaining soil quality and fertility of soil developed from Pleistocene eolian sand depositsarises from improper management and conservation of the highly fragile sandy soil environment. The researchwas undertaken to determine how various aspects of cultivated soil fertility can be related to specific pedologicaland geological factors, and which one of these relationships plays an essential role in Lumbarda “polje”, KorculaIsland, Croatia. Soil survey was done by examining and sampling one hundred locations to the depth of 100 cmwithin 40 ha of the winegrowing site. Geostatistical analysis was applied to characterize the spatial variabilityand produce the soil map of the area. Soil profiles were sampled for laboratory analysis, including physico-chemical characteristics and mineralogy. Modal analysis and petrographic microscopy of soil concretions wereconducted as well. Two soil units (SU) were identified, both containing more than 80% sand: SU1 HypoluvicArenosols, and SU2 Haplic Arenosols (FAO 2006). In SU1, reddish brown color originated from Fe oxide coatingson sand grains, whereas SU2 is characterized by the rise of pH with depth and the presence of calcite as acementing material. Cation exchange capacity was significantly higher in SU2. The documented variability ofelement content and distribution in the soil profiles indicates different weathering stages of the sandy layers. Di-agrams of Cr/V and Co/V ratios indicate different origin of sand deposits. Light mineral fraction is dominant inboth SU. Most of the sand grains are intensively weathered and well rounded, indicating long transportationfrom the source area. The results of mineralogical and geochemical characterization of Lumbarda “polje” grape-vine growing site are applicable and useful to future studies that involve “fingerprinting” regional wines.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Spatial variability of soils is a natural consequence of soil formingprocesses, but in the examination of agricultural land, considerable ad-ditional variability appears as a result of long-time specific cultivationpractices. Sandy soils are particularly sensitive to those variability fac-tors. However, the presence of paleosands on several locations alongthe Eastern Adriatic coast has not been previously subjected to detailedsedimentological and pedogenic studies. The most recent studies thatfocused on the interpretation of depositional mechanisms of sand inthe archipelago of mid-Adriatic islands showed the presence of eoliandeposits intercalatedwith alluvial deposits (Pavelic et al., 2011). Despitethe fact that Pleistocene eolian sand deposits cover rather small areas,they are extremely important for viticulture and the cultural landscapeformed by agricultural evolution (Agostini et al., 2006).

Coastal eolian dunes occur throughout the world and are character-ized by a great ecological diversity in terms of geomorphological dimen-sions, environmental heterogeneity and species variability (Martinez

85 12394099.

ghts reserved.

and Psuty, 2004). However, the problems arising from their impropermanagement and conservation are greatly similar worldwide, includingwind erosion, salt exposure, drought, and nutrient deficiency (Gilbertet al., 2008; Hesp and Martinez, 2007). The role of organic carbon as akey factor of soil fertility and vegetation production has been well doc-umented previously (Dawson and Smith, 2007; Feng et al., 2002), andsoil organic carbon (SOC) is also considered to be a key index in theevaluation of soil degradation and soil C sequestration. Degradation ofsandy soils is primarily caused by a decreasing amount of SOC and isalso reflected in the deterioration of major soil functions related tophysical stability and water and nutrient storage and fluxes. Novaraet al. (2011) studied soil organic carbon dynamics in a Mediterraneansemiarid environment and pointed out that land-use conversion, vege-tation type, and management practices, including viticulture, whichcontrol the biogeochemical and physical properties of soil could helpto sequester SOC.

Lumbarda “polje” is an atypical karst depression (“polje” or field)near the city of Lumbarda on the Island of Korcula along the southernAdriatic coast. The term “polje” refers to a unique landform feature, alarge closed karstic depression in the Dinaric region along the east-ern Adriatic region and farther inland. Many of these karst fields

282 R. M. et al. / Catena 113 (2014) 281–291

are vine-growing sites. Likewise, the Lumbarda “polje” region is fa-mous for producing Grk wine from a unique white grape variety in-digenous to this particular site. In addition, individual vine plantsare extensively grown on the southern Adriatic islands. Nowadays,special attention is being paid in preserving autochthonous grape-vine cultivars and producing wines that reflect the characteristicsof the specific site environment (Zdunic et al., 2008). Archeologicalevidence suggests that the tradition of wine-growing in the regiongoes back to the 4th century B.C. Over the last decade, winegrowersfrom the Lumbarda “polje” region have encountered the problem ofmaintaining the grapevine and wine quality due to rapid deteriora-tion of soil quality (M. Romic et al., 2012).

In general, multidisciplinary approaches that integrate soil-morphological, chemical, micromorphological and mineralogical studiesof paleosols are needed to highlight soil features and properties im-portant for the sustention of soil functions (Magliulo et al., 2006).The success of agricultural production, including specific sustainablegrapevine growing, depends on interrelationships of different pedo-logical factors. Soil texture and porosity are the main factors for soilmoisture management and water holding capacity. On the otherhand, soil mineralogy and chemistry, as well as clay mineralogy,are important for supplying adequate levels of macro- and microele-ments. Reis et al. (2006) considered three types of chemical elementsfor assessing the geochemical quality of soils for grapevine cultivation:1) nutrients (P, K, Ca, Mg, Fe, B, and Mn); 2) toxic elements (Pb, Ni, Cr,Cd, As and V); and 3) elements that are potentially toxic in high concen-trations (Co, Cu and Zn).

Preliminary soil survey results in the studied area showed that theLumbarda vineyard terrains are characterized by low fertility, low con-tent of organic matter and good permeability (Bragato et al., 2009). De-pletion of soil organic matter has been intensified by sand texture, lackof cover crops and pruning wood restitutions, and unsuitable tillagepractices. As a consequence, mineral fertilizers have almost entirelysubstituted manure, whereas soil structural stability and water holdingcapacity have decreased.

The pronounced variability assessed in surface vineyard soils of theLumbarda region requires interpretations of the observed differencesfrom the geological and pedological points of view. Therefore, we de-scribed physical, chemical and micromorphological features of thepaleosols of Lumbarda in order to identify the most common attributesof their genesis.Wewere especially interested to determine howvariousaspects of soil fertility can be related to specific pedological and geologi-cal factors, andwhich one of these relationships plays an essential role inLumbarda terroir. Thus, our research was focused on: (i) the descriptionof physical, chemical, geochemical andmineralogical features of the soilsand paleosols of Lumbarda; (ii) the spatial variability of soil propertiesand the distribution of soil nutrient and tracemetal contents at a very de-tailed scale; and, (iii) the evaluation of inherent quality of paleosands ofthe Lumbarda “polje” for grapevine cultivation.

2. Materials and methods

2.1. Study area

The survey was carried out in Lumbarda “polje” (42°55′10″ N,17°11′00″ E), which is an atypical karst depression (“polje”) of asym-metrical shape located at an elevation of up to 20 m above sea level inthe easternmost part of the Island of Korcula (Croatia). Lumbarda“polje” is surrounded by Upper Cretaceous (Senonian) rudist limestonewith sporadic thin intercalations and lenses of dolomite. Limestone isfine-grained calcilutite, containing skeletal fragments (bioclasts), andbiocalcarenite with high content of calcium carbonate. According tothe Basic Geological Map of Croatia (Korcula sheet) (Korolija et al.,1977), Lumbarda “polje” was filled by Quaternary eolian sands thatoriginated as coastal paleodunes. Like many other karst fields of theAdriatic coast, the 40 ha of Lumbarda “polje” is almost completely

cultivated with grapevine — a renowned local white berry varietynamed Grk.

Meteorological data of the Korcula weather station indicate mildtemperatures in winter, with values usually higher than 6 °C, andmean daily temperatures higher than 20 °C from June to September,with a mean annual temperature of 16.6 °C. A peculiarity of Korcula isits rainfall regime. Eastern inland relieves are responsible for a highmean annual rainfall of 1000 mm, which is characterized by an unevendistribution of rain events, with a dry period from mid-June to mid-August.

2.2. Soil survey

An intensive soil survey at the 1:5000 map scale was conducted inSeptember 2008 to assess Lumbarda “polje” soil variability and its rela-tionship to Grk production. One hundred sampling locations were se-lected according to the spatial coverage design (Brus et al., 2003).Pixel coordinates were extracted from the raster image of aerial photo-graphs and processed with the k-means clustering algorithm. Theresulting centroids gave the set of coordinates of sampling locationsshown in Fig. 1.

Sampling sites were located in the field by GPS navigation and eachlocation was augered to a depth of 100 cm, recording the successionand thickness of soil horizons. Color of soil using Munsell Soil ColorCharts (1994), USDA texture class (Soil Survey Division Staff, 1993),and effervescence to 1 N HCl were determined. Soil horizons werethen sampled, air-dried, sieved at 2 mm and analyzed in the laboratoryfor grain size distribution (dry sieving + pipette method) and pH in a1:5 soilweight/water volume ratio (Dane andTopp, 2002). Themultivar-iate set of quantitative and qualitative data summarized in Table 1 wasanalyzed to determine pairwise similarities according to Goodall's simi-larity index approach (Goodall, 1966; Goodall et al., 1987). The resultingsimilarity matrix was partitioned into homogeneous groups by hierar-chical cluster analysis using several clustering criteria (Podani, 2000).Vectors corresponding to observations with the characteristics of eachgroup centroid were extracted from the similarity matrix and subjectedto geostatistical analysis to characterize the spatial variability of eachgroup of observations and to draw a draft version of the soil map of thearea.

2.3. Soil profiles and physicochemical analysis

Ten auger samples per soil unit were randomly selected to charac-terize the physico-chemical characteristics of Ap and C horizon samples.Two soil profiles per unit (Fig. 1) were then located according to the soilmap, dug down to the parentmaterial (140–200 cm depth), and exam-ined morphologically. Soil horizons were sampled for laboratory analy-ses and mineralogical characterization.

Soil samples were air-dried, sieved at 2-mm, and subjected to thefollowing analyses: pH in a 1:5 soil/water ratio (MettlerToledo MPC227 pH-meter), soil organic carbon (SOC) by sulfochromic oxidation(HRN ISO 14235, 1998), calcium carbonate (CaCO3) by the volumetriccalcimeter method after HCl attack, available K and P by the ammoniumlactatemethod (Egner et al., 1960), and effective cation exchange capac-ity (CEC) using BaCl2 solution. Soil samples were also digested in aquaregia (HRN ISO 11466, 2004) with the microwave technique on aMARSXpress system (CEM). Element concentrations in soil digests (Al,Ca, Co, Cr, Fe,Mg,Mn,Ni, P, Pb, V and Zn)were determinedby inductive-ly coupled plasma optical emission spectroscopy (ICP-OES) on a VistaMPX AX (Varian). All concentrations were calculated on the basis ofdry weight of samples (105 °C, 24 h).

Quality control procedure consisted of reagent blanks, duplicatesamples and several referenced soil and sediment samples with similarmatrix from the inter-laboratory calibration program (Houba et al.,1996). Maximum allowable relative standard deviation between repli-cates was set to 10%.

Fig. 1. Aerial photograph of the study area and sampling locations in Lumbarda “polje” on the Island of Korcula (southern coast of the Adriatic Sea, Croatia).

283R. M. et al. / Catena 113 (2014) 281–291

2.4.Modal analyses of soil samples and petrographicmicroscopy of carbonateconcretions

Mineral composition of soil horizon samples was determined bymeans of analysis of thin sections using a polarization microscope.Modal analysis of soil horizon samples was performed in the0.09–0.16 mm calcite-free fraction. Heavy and light mineral fractions(HMF and LMF) were separated by bromoform liquid (δ(CHBr3) =2.85–2.88 g cm−3). Qualitative and quantitative analyses of HMF andLMF for 11 samples were performed by identifying 300–400 grainsper sample using polarizing light microscope (Mange and Maurer,1992).

Petrographic thin sections were also prepared from concretionsfoundwithin the soil. They were investigated under a polarizing micro-scope to determine mineral composition of the concretions and to doc-ument the presence of any voids and cracks and their infilling and/orcoating, which could provide information on the mineral compositionand origin of parent material.

Table 1Descriptive statistics of C horizon attributes used for Goodall's similarity index determina-tion (n = 100). The frequency of occurrence fi of each discrete state is reported for binaryand ordinal attributes.

A. Discrete attributes

Attribute Type State fi (%) State fi (%) State fi (%)

Munsell value Binary 3/ 80 4/ 20Munsell chroma Binary /3 30 /4 64 /6 5Effervescence Binary Absent 76 Weak 24Texture classa Ordinal SL 11 LS 43 S 46

B. Continuous attributes

Attribute Mean Std. Dev. Min Max

Sand, g kg−1 860 60 680 970D50 sand, μm 222 13 188 248pH (1:5 soil/water) 7.02 0.66 5.52 8.44

a SL: sandy loam; LS: loamy sand; S: sand.

2.5. Statistical analysis

Mean and standard error of selected soil chemical characteristicsand trace metals were calculated and two-way ANOVA F-test was ap-plied considering the horizon type and soil unit as factors of variation.The statistical analyses were conducted using the Statistical AnalysisSoftware (SAS 9.1).

3. Results and interpretations

3.1. Environmental setting and soil classification

Before the determination of Goodall's similarity index from themul-tivariate set of data, an explorative geostatistical analysis was donewiththe quantitative attributes reported in Table 1. The pH of C horizonshowed the largest variation, from 5.52 to 8.44, and resulted in a well-structured variogram and a spatial pattern related to the morphologyof soil surfaces. Values larger than 7.40, in particular, were present inthe central part of Lumbarda “polje”, characterized by a morphologicaldepression surrounded by limestone outcrops, and in its eastern por-tion, which is in contact with a terrace escarpment exposing carbona-ceous sandy strata (Fig. 2).

The attributes of Table 1, the presence of still recognizable geogenicsignatures and the meteorological data from the Korcula weather sta-tion allowed us to classify Lumbarda soils as Arenosols according to theWorld Reference Base (FAO/ISRIC/ISSS, 2006). The combination ofGoodall's similarity index and hierarchical clustering suggested afurther distinction into two groups of soil observations, with theircentroidal values reported in Table 2. Centroids were equivalent interms of color, texture class and effervescence, but differed in pH and,to a lesser extent, sand content.

The similarity vector of the soil observation nearest to Cluster 2 cen-troid was processed with geostatistics and its interpolation map wasdefuzzified to delineate the two soil units by assigning full membershipof grid nodes with similarity values larger than 0.6 to soil unit 2 (SU2),and full membership to SU1 in other cases. The soil units were classifiedas SU1—Hypoluvic Arenosols, and SU2—Halpic Arenosols, and their spa-tial distribution is reported in Fig. 3.

Fig. 2. Spatial distribution of horizon C pH in the study area.

284 R. M. et al. / Catena 113 (2014) 281–291

3.2. Physical and chemical data of selected augerings

Soil units were further characterized by analyzing horizon samplesof 10 randomly-selected boreholes per unit. Table 3 reports summarystatistics of several soil characteristics and the results of two-wayANOVA F test considering the horizon type and soil unit as factors ofvariation.

Several soil properties show significant differences between Ap andC horizons, which are, however, not attributable to the same environ-mental factor. Sand and silt variation with depth is probably related tocoastal sedimentation processes. The hypothesis that the variationmay be determined by alteration of sands and their disintegration intosilt-sized particleswould have to be tested bymore detailed sedimento-logical investigations.

The decrease of SOC, P2O5 and K2O and Kex – the latter being charac-terized by a Pearson's r of 0.94 – is on the other hand related to fertiliza-tion practices in Lumbarda “polje”. The amounts of SOC and P2O5 arevery low in Ap horizon. The behavior of SOC, in particular, can beexplained in termsof depletion rather than accumulation in soil becauselivestock breeding on the island has been drastically reduced in the lastdecades, consistently decreasing the amendment of vineyards with cat-tlemanure. In Lumbarda “polje”, the depletion of soil organicmatter hasbeen accentuated by sand texture, the lack of cover crops and pruningwood restitutions, and by unsuitable tillage practices. As a consequence,mineral fertilizers almost entirely substitutedmanure,whichmay resultin decreasing soil water holding capacity.

Significant differences between soil units are instead recorded forCEC, which is significantly higher in SU2 unit. Even if not significant,CEC also shows an increase from Ap to C horizon that is related to thelarger content of silt, hence the increase of exchange surfaces in thesoil. CEC is fully saturated with exchangeable cations whose content

Table 2Centroidal values of SU1 and SU2 clusters from augering observations.

Cluster Munsellvalue/chroma

Texture class Effervescence Sand,g kg−1

D50 sand,μm

pH

1 3/4 Loamy sand Absent 810 233 6.762 3/4 Loamy sand Absent 850 226 7.34

decreases in following order: Ca–Mg–K–Na. Even if onlyMg significant-ly increases from SU1 to SU2, its association with a slight increase in Casuggests a relationship between CEC and the origin of parent material.

Concentrations of all soil trace metals in the study area are lowerthan the median Coastal Croatia background value, except for Cu andNi (Table 4). A statistical difference in tracemetal contentwas observedbetween soil units, with the exception of Cu and Pb (Table 4). No statis-tical difference between soil unitswas observed for phosphorus and sul-phur content. A statistically significant difference in Cu, S and P contentbetween different soil horizons was observed. High concentrations ofthese elements in the surface horizonwere directly related to the appli-cation of agrochemicals in grapevine growing.

3.3. Physical and chemical properties of soil profiles

Table 5 reports selected physical and chemical analytical featuresof the four soil profiles, profiles 1 and 4 belonging to SU1 unit, andprofiles 2 and 3 to SU2 unit. The data are in agreement with thosepresented in Table 4. In particular, it is noticeable that the increasein both CaO and MgO in SU2 profiles is in accordance with the in-crease of Caex and Mgex.

Profiles P-1 and P-4 are characterized by an Ap–CI–CII and Ap–Bt/CI–CII horizon sequence, respectively, and a reddish brown colororiginated from Fe oxide coatings on sand grains (Fig. 4), whereasprofiles P-2 and P-3 display a rise of pH and CaO with depth, whichis related to local concentrations of calcite as a cementing material(Fig. 5). Compared to P-2 and P-3, profiles P-1 and P-4 show finergrained texture – loamy sand to sandy loam vs. sand to loamy sand –

and a slightly higher SOC content and available nutrients. A specific fea-ture of these profiles is a sharp increase in clay content that character-izes CII and Bt/CI horizons of P-1 and P-4 profiles, respectively. Thislithological discontinuity is associated with the highest Fe2O3, Al2O3

and K2O content recorded in these profiles, suggesting the presence ofilluviation stratum that is usually found closer to the surface in semiaridand arid climates where precipitation is scarce.

Trace metal contents in the studied soil profiles are presented inTable 6. The variability of both elemental content and vertical distri-bution is highly pronounced and indicates different weatheringstages of the sandy layers. In general, two different patterns of ele-mental distribution within the profiles were observed. In P-1 and

Fig. 3. Spatial distribution of the two soil units delineated in the study area.

Table 4

285R. M. et al. / Catena 113 (2014) 281–291

P-4 profiles put in SU1 elemental concentrations significantly in-crease with depth. In SU2 profiles P-2 and P-3, the fine sand fractiondominates and pH rises with depth, but elemental concentrations,except for Ca and Mg, decrease. Some trace metal concentrations inthe studied soils, like Cr and Ni, are considerably higher than the me-dian for Coastal Croatia (Halamic and Miko, 2009). It can be seenfrom Table 6 that Al, Co, Cr, Fe, Mn, Ni, V and Zn show similar behav-ior in the two distinguished soil units. Concentrations of those ele-ments are higher in SU2 with carbonate matrix within the upperhorizon and decrease with depth. Vertical element distributionswithin SU1 profiles display a reverse pattern, becoming generallyhigher in deeper horizons compared to SU2. In P-1 and P-4, Ca andMg contents are rather uniform within the profile, but in P-2 and P-3 their contents increase considerably with depth. The distributions

Table 3Mean, standard error (in parentheses) and two-wayANOVA F test of selected soil chemicalcharacteristics (ni = 10).

Variable Horizon Soil unit Pr(F)

SU1 SU2 Horizon Soil unit

Sand, g kg−1 Ap 913 (12.0) 904 (19.0) N0.01 n.s.C 866 (18.0) 845 (21.0)

Silt, g kg−1 Ap 51.0 (7.0) 59.0 (12.0) N0.01 n.s.C 91.0 (11.0) 112 (18.0)

Clay, g kg−1 Ap 36.0 (6.0) 38.0 (7.0) n.s. n.s.C 42.0 (10.0) 44.0 (7.0)

OC, g kg−1 Ap 5.72 (0.55) 8.14 (0.73) N0.01 n.s.C 4.14 (0.54) 4.97 (0.56)

pH Ap 6.86 (0.19) 7.25 (0.23) n.s. n.s.C 7.10 (0.25) 7.00 (0.18)

CEC, cmol(+) kg−1 Ap 5.00 (0.63) 7.44 (0.79) n.s. N0.01C 5.65 (0.63) 7.97 (0.82)

Caex, cmol(+) kg−1 Ap 4.54 (0.65) 6.06 (0.74) n.s. n.s.C 5.35 (0.60) 6.80 (0.77)

Kex, cmol(+) kg−1 Ap 0.58 (0.06) 0.83 (0.15) N0.01 n.s.C 0.41 (0.06) 0.49 (0.06)

Mgex, cmol(+) kg−1 Ap 0.78 (0.12) 1.28 (0.14) n.s. N0.01C 0.78 (0.13) 1.27 (0.15)

P2O5, mg kg−1 Ap 115 (22.1) 93.6 (13.3) N0.01 n.s.C 62.5 (15.6) 56.8 (10.2)

K2O, mg kg−1 Ap 269 (34.2) 329 (42.9) N0.01 n.s.C 181 (26.4) 219 (29.6)

of Cu, P and S in soil profiles indicate enrichment in the upper hori-zon due to agricultural land use, as already noticed from Table 4.

In profile P-1, the highest concentration of Al, Cr, Fe, Mn, Ni and Vwas identified at depth of 115–135 cm. This corresponds to a substan-tial increase in clay content in CII horizon of this profile. These elementsare probably part of layer-silicate ormetal oxide structural components.Similar features are noticed in profile P-4, where an argic horizon with150 g kg−1 of clay is present at a depth of 70–95 cm. The clay contentdecreases to 90 g kg−1 in the underlaying CII horizon. Although profilesP-1 and P-4 belong to the same soil unit, the differences in their featuresmay be assigned to soil translocation within the landscape by tillage.

Mean, standard error (in parentheses) and two-way ANOVA F test of trace metals in soil(ni = 10).

Variable Horizon Soil unit Pr(F) CoastalCroatiaa

SU1 SU2 Horizon Soil unit

Al, g kg−1 Ap 19.3 (2.0) 23.5 (1.5) n.s. N0.05 78.5C 20.2 (1.5) 24.6 (2.0)

Co, mg kg−1 Ap 10.1 (0.8) 14.4 (0.9) n.s. N0.01 18C 11.0 (0.7) 14.5 (1.2)

Cr, mg kg−1 Ap 70.3 (6.7) 99.6 (6.9) n.s. N0.01 121C 73.2 (4.5) 99.4 (8.3)

Cu, mg kg−1 Ap 132 (17.5) 134 (11.0) N0.05 n.s. 35.5C 102 (11.7) 101 (9.3)

Fe, g kg−1 Ap 21.3 (1.8) 27.4 (1.7) n.s. N0.01 41.8C 23.0 (1.5) 28.0 (2.2)

Mn, mg kg−1 Ap 458 (42.6) 727 (54.0) n.s. N0.01 1082C 547 (59.6) 719 (67.8)

Ni, mg kg−1 A 79.7 (9.7) 126 (12.3) n.s. N0.01 74.6C 85.5 (8.3) 132 (14.4)

P, mg kg−1 Ap 379 (41.4) 394 (38.9) N0.01 n.s. 650C 291 (21.4) 292 (15.8)

Pb, mg kg−1 Ap 12.1 (1.0) 12.7 (1.0) n.s. n.s. 48.7C 13.6 (1.1) 12.5 (1.1)

S, mg kg−1 Ap 117 (9.2) 137 (8.3) N0.01 n.s. –

C 92.7 (5.7) 104 (6.0)V, mg kg−1 Ap 47.9 (4.0) 59.8 (3.4) n.s. N0.01 148

C 50.1 (3.1) 60.3 (4.2)Zn, mg kg−1 Ap 40.8 (2.5) 48.2 (1.7) n.s. N0.01 108

C 41.1 (1.6) 45.9 (2.5)

a Median, after Halamic and Miko (2009).

Table 5Selected physical and chemical properties of examined soil profiles.

Soil unit Profile Horizon Depth Munsel color Sand Silt Clay pH K2O CaO Na2O MgO Al2O3 Fe2O3 P2O5 MnO

cm Dry g kg−1 H2O g kg−1

SU1 P1 Ap 0–60 5 YR 4/4 820 130 50 7.17 3.80 2.90 0.20 8.30 35.3 31.1 0.70 0.50CI 60–115 5 YR 5/4 810 150 40 8.03 3.80 2.70 0.20 8.20 33.2 32.6 0.30 0.50CII 115–135 5 YR 4/6 670 180 150 7.23 6.00 3.10 0.40 10.9 61.3 51.9 0.50 1.00

P4 Ap 0–70 7.5 YR 5/4 880 120 0 8.54 3.00 8.20 0.30 7.6 19.3 21.2 0.40 0.40Bt/CI 70–95 7.5 YR 4/4 720 130 150 8.18 6.10 4.40 0.30 10.6 50.4 48.5 0.60 1.00CII 95–205 7.5 YR 4/4 800 110 90 8.04 4.80 3.10 0.30 10.7 37.3 39.9 0.40 0.80

SU2 P2 Ap 0–40 7.5 YR 4/4 790 160 50 7.93 6.20 6.60 0.60 15.9 46.3 49.3 0.40 1.10CI 40–90 10 YR 4/3 930 70 0 8.96 4.40 3.26 0.60 35.5 38.8 43.9 0.40 1.30CII 90–180 10 YR 5/3 910 40 0 9.09 2.50 144 0.50 44.9 23.2 31.0 0.40 0.10

P3 Ap 0–75 7.5 YR 4/4 880 110 10 8.10 3.10 5.90 0.40 25.7 31.4 36.7 0.30 0.90C 75–140 2.5 YR 4/4 940 60 0 9.00 2.00 130 0.50 42.0 25.5 28.8 0.30 0.70

286 R. M. et al. / Catena 113 (2014) 281–291

Soilmovement by tillage can be the dominant force in redistributing soilwithin the profile and throughout the landscape (Papiernik et al., 2007).In spite of the fact that the parcel where profile P-1 was located hadbeen abandoned a long time ago and is not used for grapevine growinganymore, the highest Cu content of 127 mg kg−1 was determined in itssurface layer.

Chromium content is much higher in SU2 profiles, approachingthe median value for soil Cr in the coastal Croatia region of121 mg kg−1. Nickel concentrations are also much higher in SU2than in SU1, being many fold higher than the baseline for soil ofcoastal Croatia (74.6 mg kg−1).

3.4. Soil mineralogy

The results of modal analysis of soil samples indicated the dom-inance of light mineral fractions (LMFs) relative to heavy minerals,which constitute less than 20% in SU1 profiles and about 30% inSU2 profiles (Table 7). Quartz and transparent lithic particlesmake up approximately 90% of the light fraction. Feldspars aremostly colorless and weathered, sometimes kaolinized. WeatheredK-feldspars occur as monocrystals with zircon inclusions, whilefresh grains are less common and mostly represented by orthoclaseand rare microcline.

The abundance of lithic particles is greatest in SU2 profiles. Lith-ic particles are mainly well rounded quartzite grains, with some“impurities” of sericite, chlorite, glaucophane, hornblende, biotite,or their combinations. If the grains are parts of schistose rocks,they are most commonly fine-grained black shale, greenschists (presentin all samples) or blueschist. In the LMFmuscovite is rare and exclusivelyin the form of sericite. Other lithic particles are mostly quartzite grains

Fig. 4. Field and petrographic observations of concretions from profile P-1 (SU1). (a) Iron oxide c(115–135 cm). Fe/Mn-oxides and hydroxides and/or clay minerals are dispersed within concretiowalls of cracks. In the right upper corner there are voids filled with organic matter (arrows). Plain

containing chlorite, glaucophane, hornblende, biotite and epidote orzoisite.

Heavy mineral fraction was relatively higher in profiles from SU2compared to SU1 profiles. In the heavy mineral fraction, transparentheavy minerals representedmore than 80% in SU1 profiles, andmorethan 90% in SU2 profiles. Somewhat greater amount of micaceousminerals biotite and chlorite is present in SU2 profiles. The mostabundant transparent heavyminerals are amphiboles and pyroxenesfrom the epidote–zoisite mineral group. This mineral group is repre-sented by equidimensional, irregular, weathered grains. Epidote isusually yellow to greenish yellow and shows weak pleochroism,while zoisite is colorless and with characteristic anomalous blue in-terference color. Amphiboles are represented by hornblende-typeamphiboles, tremolite and glaucophane. These grains are elongatedand do not always have pronounced cleavage and pleochroism be-cause of alteration products. Hornblende-type amphiboles of variouscolors (green–olive green, pale brown–brown or dark greenishbrown) are the most abundant amphiboles. Glaucophane amphiboles(pleochroic and bluish–purple–colorless) are very rare. Pyroxenes con-sist mostly of orthopyroxene, whereas in P-3 at depth 75–140 cmclinopyroxenes are dominant. The amount of epidote–zoisite dominant-ly increaseswith depth in SU1 soil profiles, while amphiboles and pyrox-enes decrease with depth. The amount of amphiboles and pyroxenes islower within SU1 profiles in comparison with SU2 profiles. Chromitesare sporadic,mostly dark reddish brown, present in SU1profiles togetherwith a very small portion (b3%) of ultra stable minerals zircon, tourma-line and rutile.

Generally, the abundance of heavy mineral fraction is higher in SU2,while the amount of transparent heavy minerals in this soil unit isslightly lower than in SU1. In SU1, the percentage of epidote–zoisite in-creases with depth, and the amount of amphiboles and pyroxenes

oatings on sand grains in profile P-1. (b) Photomicrograph of a siliceous concretion from P-1ns as groundmass or form brown coatings with laminae parallel to the surfaces of grains orpolarized light.

Fig. 5. Field and petrographic observations of carbonate concretions from profile P-3. (a) Field photograph of concretions elongated parallel to cross-laminated eolian deposits. (b) Pho-tomicrograph of carbonate concretion. Fine crystalline equant calcite cement occludes the interparticle pores. Detritalminerals aremostly carbonate grains, rarely quartz or quartzite. Plainpolarized light.

287R. M. et al. / Catena 113 (2014) 281–291

decreases with depth, while the distribution of those minerals in SU2 israndom. The entire sample (LMF and HMF) reflects the contribution ofindividual transparent heavy minerals, which significantly differs be-tween SU1 and SU2.

3.5. Petrographic observations of concretions of the studied soils

A SU1 concretion and its photomicrograph from CII horizon of profileP1 are shown in Fig. 4a and b, respectively, and a SU2 concretion and itsphotomicrograph fromCI horizon of profile P3 are shown in Fig. 5a andb,respectively. SU1 concretions are postpedogenic siliciclastic concretionslocated in the base of recent soil cemented by fine grained brown coat-ings. Grains within the concretions aremostly quartzite and chert, rarelyquartz and K-feldspars (orthoclase and scarcely microcline), plagioclase,light green brown amphiboles and epidote. Fine-grained chlorite is dis-persed within feldspar grains. Grains are subangular to round. Clay min-erals and Fe/Mn oxides and hydroxides are dispersed in concretions as agroundmass or form brown coatings with laminae parallel to the sur-faces of grains or walls of cracks and voids.

SU2 concretions are postsedimentary carbonate concretions precip-itated in coarse-grained laminae of laminated sediment. Carbonate con-cretions do not form continuous layer (Fig. 5a). They are composedmainly of carbonate grains, rarely quartz, quartzite, green hornblendeand some intensively weathered grains. Scarcely, there is some chlorite

Table 6Trace elements content in soil profiles.

Soilunit

Profile Horizon Depth Co Cr

cm mg kg−1

SU1 P1 Ap 0–60 10.9 7CI 60–115 12.5 7CII 115–135 17.2 10

P4 Ap 0–70 7.79 5Bt/CI 70–95 17.5 10CII 95–205 16.1 9

SU2 P2 Ap 0–40 19.4 12CI 40–90 18.0 13CII 90–180 12.3 8

P3 Ap 0–75 15.5 10C 75–140 12.7 8

as a product of alteration. Matrix is fine-grained calcite (microsparite).Mineral composition of these concretions indicates that they precipitat-ed before pedogenesis took place and cementation preserved themfrom pedogenetic influence.

4. Discussion

Soil types developed on different substrates are likely to befingerprinted by distinctive geochemical composition. Subsequently,the concept of viticultural terroir is based on the idea that unique char-acteristics of a certain plot of land and its slope contribute to the distinc-tive character of a wine and that the character given to the wine fromthe soil is different for the same grape grown on different plots of land(White, 2003). Although the pedogeochemical study of viticulturalsoils of the Lumbarda region has not been targeted at the identificationof sediment origin, the results clearly demonstrate the variability of thestudied soil environment.

The two soil units delineated in the study area belong to differentsoil types, in SU1 — Hypoluvic Arenosols, and in SU2 — Haplic Arenosols,though highly antropogenized. This allows identification of the twodominant and strongly linked soil factors. First is the different parentmaterials (carbonate vs. silicate), and the other is clay illuviationresulted in the presence of an argic horizon in silicate material. Theinfluence of the first factor is outlined by the bulk soil data and metal

Cu Ni Pb V Zn

4.9 128 105 13.9 51.8 41.48.7 11.9 114 9.02 51.5 35.79 15.1 204 10.7 78.3 48.41.0 71.1 61.3 7.40 34.2 37.24 15.8 205 11.9 74.3 46.31.3 12.5 178 8.96 63.0 42.09 30.4 236 9.20 71.0 47.93 11.5 242 5.69 63.4 45.61.1 7.31 136 5.09 39.4 30.48 40.5 193 9.43 48.9 36.16.9 7.98 144 6.02 41.8 29.1

Table 7Modal composition of the heavy and light mineral fractions (0.09–0.16 mm) of soil profile concretions from Lumbarda site. LMF— light mineral fraction, HMF — heavy mineral fraction,THM— transparent heavyminerals, q— quartz, f— feldspar,m—muscovite, l— transparent lithic particles, op— opâqueminerals, ch— chlorite, b— biotite, ep.zt— epidote–zoisite, am—

amphibole, py— pyroxene, g— garnet, cy— kyanite, st— staurolite, tu— tourmaline, zr— zircon, ru— rutile, ti— titanite, ap— apatite, cr— chromite,+—mineralswith occurrence b 0.5%.

Profile Depth (cm) Composition of LMF100%

HMF % Composition of HMF100%

THM 100%

q f m l op ch b THM ep zt am py g cy st tu zr ru ti ap cr

1 Ap 0–60 34 11 + 55 16.2 2 2 4 92 16 33 45 0 + 2 1 2 0 1 0 +C I 60–115 37 8 0 55 15.5 + + 5 94 25 30 43 1 0 0 0 0 0 0 0 0C II 115–135 42 12 0 46 8.68 4 2 5 88 44 27 27 0 0 0 + + + + + +

2 Ap 0–40 31 6 0 63 22.2 2 6 6 86 38 32 30 0 0 + 0 0 0 0 0 0C I 40–90 38 10 0 52 37.5 1 6 3 90 24 44 31 0 0 1 0 0 0 0 0 0C II 90–180 65 11 + 23 38.2 + 9 10 82 27 43 29 0 0 1 1 0 0 0 0 0

3 Ap 0–75 26 3 0 71 27.4 0 6 13 82 24 46 30 + 0 0 0 0 0 0 0 0C I 75–140 35 7 0 58 29.8 + 7 9 84 23 37 38 0 + 0 1 0 0 + + 0

4 Ap 0–70 39 6 0 55 18.5 1 1 2 95 22 40 36 0 0 + 0 + + 0 0 +Bt/CI 70–95 40 10 0 50 8.51 + 2 3 96 37 34 27 0 0 0 + 0 + 0 0 +C II 95–205 34 8 0 58 11.3 + 2 2 96 41 31 28 0 0 0 + 0 0 0 0 0

288 R. M. et al. / Catena 113 (2014) 281–291

concentrations, and the impact of the second factor is reflected in themodal analysis that relies on thin section petrography to define thepresence of the argic horizon.

Recent studies of Pleistocene deposits of the eastern Adriatic coast(Pavelic et al., 2011) concluded that eolian deposits are intercalatedwith alluvial deposits. Present-day characteristics of soil depend onmany factors, startingwith depositional mechanisms and environment,paleogeography and characteristics of the sedimentary basin, and con-tinuing during soil formation and pedogenesis. All these factors play aparticular role in determining the geochemical composition of soilsthrough selective element enrichment or depletion. Amorosi andSammartino (2007) highlighted a great use of facies analysis and accu-rate reconstruction of historical changes in drainage patterns in under-standing distinctive patterns of elements, especially metal distributionin soils.

Modal analyses indicate that mineral composition of soils in thestudy area is specific. Most of the grains are intensively weathered andwell rounded, indicating long transportation from the source area toLumbarda, where they were deposited and exposed to pedogenesis.Considering the composition of LMF (dominated by lithic particles,mostly quartzite and shale) and HMF (almost exclusively amphiboles,pyroxenes and epidote–zoisite group of minerals), as well as the highportion of HMF (Table 7), the only possible source area is the drainagebasin of the upper Neretva River. The sediment source is the DinarideOphiolite Zone with ophiolites and ophiolite mélange composed ofrocks with compositions that perfectly correspond to those of analyzedsamples from Lumbarda (namely the variety of amphibolites, gabbro,diabase–dolerites, chert, shales, greenschist, graywacke, basalts)(Pamic et al., 2002; Pavelic et al., 2011; Robertson et al., 2009). Thesezones are rich in heavymetals, especially Cr, Fe and V. Since the NeretvaRiver was transporting material through the Dinaride carbonate plat-form, during transportation thatmaterial could have also been enrichedwith Paleogene bauxites from the surrounding deposits that may con-tain approximately 1.100 mg kg−1 of Cr (Kovacevic Galovic et al.,2012; Pamic et al., 2002).

After reaching the Adriatic coast, the sediment was dispersed by al-luvial processes and later redeposited by wind. These eolian sedimentswere deposited and preserved on the present-day Adriatic islands. Inthe Late Pleistocene along the Eastern Adriatic coast, the sea-level waslower than −23 m during MIS 7b (202 ka BP), above −14 m duringthe MIS 5a (84 ka and 77 ka BP), and followed by rapid Holocene sea-level rise from lower than −41.5 m (9.2 ka BP), to −10 m (7.8 kaBP), and −1.5 m (3.4 ka BP) (Suric and Juracic, 2010). This is in agree-ment with the investigations of sea-level oscillations during that periodin the Adriatic basin and within the Mediterranean region (Lambecket al., 2002; Suric et al., 2009). During lower than present sea level theNeretva River and its tributaries formed alluvial plain (basin) amongthe present-day Dalmatian islands and these alluvial sediments were

subsequently redeposited by marine processes or wind (Pavelic et al.,2011). These alluvial and eolian sediments were parent material forpaleopedogeneses or recent pedogeneses of analyzed horizons inLumbarda “polje”. Because of climate changes that caused sea-level os-cillations (Suric and Juracic, 2010), several periods of sedimentation andpedogenesis took place. The results of someof themwere preserved andanalyzed in the Lumbarda area.

Themain constituents of eolian sediments in thenorthern part of theAdriatic basin are quartz and micaceous minerals (Mikulcic Pavlakovicet al., 2011). Sand grains on Susak Island are angular and the percentageof HMF is smaller than in samples analyzed from the Lumbarda area.The heavy mineral association points to metamorphic and igneousrocks from the Alpine regions as the sourcematerial for Susak and its re-lationship to the southward extension of the River Po plain (MikulcicPavlakovic et al., 2011).

Durn et al. (2007) analyzed heavy minerals in terra rossa soil fromthe western part of the island of Korcula and documented significantabundances of garnet, spinel, amphibole, epidote, orthopyroxene andFe–Ti oxides. Mineral composition from this location differs fromLumbarda indicating different sourcematerials and paleoenvironments.

Trace element concentrations in sediments result from the com-bined influences of provenance, weathering, diagenesis, sedimentsorting and aqueous geochemistry of individual elements (Rollinson,1993). Soils in Lumbarda may have become enriched in Cr, Fe, Mn andV because of their longitudinal distribution pattern. Jurina et al. (2010)described disposal of metals in the river-dominated deltaic depositionalsystem of the Neretva River and its adjacent coastal region. The highestconcentration of metals was found in the semi-enclosed area of theNeretva Channel characterized by deposition of fine-grained particlesand prevalent accumulation of metals. Fe oxide and oxyhydroxide coat-ings on claymineral surfaces act as a major factor in the adsorption anddeposition of trace elements includingheavymetals (Jurina et al., 2010).Such coatings have been observed in the study area (e.g., Fig. 4b) andthe corresponding high concentrations of heavy metals are notsurprising.

Production of grapevine generally requires regular application andfrequent repetition of agricultural management practices for a numberof years. In terms of heavy metals, accumulation of Cu, Zn and Cd isthe most common effect of fertilization and disease and pest control invineyards. Quantification of input of these elements through growingpractices is possible, however only as an estimate. Contribution of aerialdeposition to metal accumulation in soils is much more difficult toestimate for this region, but it is not likely or is almost impossiblethat arial deposition could by itself result in anomalous concentra-tions. High Cu concentrations are not unusual for vineyard soils.The Bordeaux mixture, an efficient agent for the prevention of vinedowny mildew, has been routinely used in Croatia since the end ofthe 19th century with the concentration and number of treatments

289R. M. et al. / Catena 113 (2014) 281–291

depending on weather conditions, infection intensity and vineyardlocation. As Cu from the Bordeaux mixture is very reactive in soil,it is found in all matrix components, and its complexation with or-ganic matter is one of the most efficient mechanisms of Cu2+ reten-tion in soil (McBride, 1994). This restricts Cu bioavailability, but alsoconsiderably reduces the risks of phytotoxicity of the accumulatedanthropogenic input and its vertical migration. Substantial reduc-tion in manure application resulted in SOM depletion thus reducingthe soil buffer capacity. Regular monitoring of soil pH during thegrowing season revealed drastic variations that may represent oneof the factors of plant stress and metal toxicity (D. Romic et al.,2012).

The main differences among the profiles examined concern thethickness of each horizon and the boundaries between them. Theplow horizon (Ap) depth varied from 40 cm to 70 cm depending onthe deep plowing depth before planting. Traditional mid-row cultiva-tion is the regularly usedmanagement practice in spite ofmany adverseeffects of constant cultivation on soil properties. Cover crops are seldomused in vineyard floormanagement. Themain reasons theGrk grape va-riety is well adapted to the extreme conditions of Lumbarda sandy soilsmay be its high vigor and deep root distribution, which improves theregularity of water supply to the vine. In a certain way, this may offera high degree of buffering against climatic extremes during longdrought periods.

The distribution of trace elements that are immobile, like Cr and Ni,and remain in the solid load during erosion and sedimentation is a use-ful indicator of geological processes and provenance (Amorosi andSammartino, 2007; Bauluz et al., 2000). These elements are believedto be transported exclusively in the terrigenous component of sedimentand therefore reflect the chemistry of sediment source (Vital andStattegger, 2000). Higher Cr and Ni are possibly related to more chloritein the fine-grained fractions (Tripathi and Rajamani, 1999). Diagrams ofCr/V and Co/V ratios shown in Fig. 6 reveal two distinct linear trends, in-dicating different provenance of sandy sediments.

The process of soil organic matter enrichment, its stabilization andretention, is much slower than the depletion. The abundance of organicand inorganic C in the soil is affected by climate and vegetation cover,and the role of organic carbon as a key factor of soil fertility and vegeta-tion production has been well documented in the literature. In generalterms, the fertility of soils of the Lumbarda region, dedicated mostly tovineyard culture for centuries, is rather low. Soil organicmatter displays

Fig. 6. Ratios between Cr and V and bet

very low values (average 5.72 g kg−1 and 8.14 g kg−1 in surface layerof SU1 and SU2, respectively, and average 4.2 g kg−1 and 4.4 g kg−1

in subsoil of SU1 and SU2, respectively) that can be explained interms of depletion rather than accumulation in soil because livestockbreeding on the island has been drastically reduced in the lastdecades, consistently decreasing the amendment of vineyards withcattle manure.

Mg deficiency, as also evident in vineyards of the study area, is com-monly associatedwith excessive use of K fertilizers. The exchangeable Kto exchangeable Mg ratio reported in Table 3 indicates the presence ofnutritional imbalance in grapevine. In particular, when the 0.5-value isexceeded, Mg-deficiency takes place, causing discoloration and earlydrop of basal leaves. In Lumbarda “polje”, a certain Kex/Mgex imbalanceis present in Ap horizons of profiles 1 and 4 and the available potassiumcontents (Table 3) suggest that this imbalance may be related to exces-sive K in soil solution. High to very high potassium content in petioles atsame plots at flowering and veraison stages (M. Romic et al., 2012) aswell as calcium and magnesium deficiency support this assumption.Different forms of soil K are present in a dynamic equilibrium and arenot all available for plant uptake (Mengel and Kirkby, 1987). Plant avail-ability of soil potassium is controlled by dynamic interactions amongthese different forms of potassium. Solution K and exchangeable K arereadily available to plants, while non-exchangeable K (fixed K andstructural K) is slowly available and makes up the main K reserve inthe soil. In addition, K availability depends on the rate of K+ uptake byroots and certain soil characteristics such as mineralogy, texture, CEC,moisture, temperature, pH, Ca, Mg and K fixation (Kirkman et al.,1994; Mpelasoka et al., 2003). As already mentioned, the fertility ofsoils in the study area is not sufficient to provide most of the nutrientsrequired, but K levels in soil solution related to the soil mineralogyseem to be excessive. Soils with low level of plant-available nutrientsof markedly seasonal character are commonly present in regions ofMediterranean climate (Milla et al., 2005). The very extensive root sys-tem of Grkmay effectively enhancemobilization and retention of nutri-ents andwater even during the long drought period. Nutrients taken upby deep roots are transported into the above-ground parts and re-deposited on the soil surface through litterfall, stemflow or throughfall(Rengel, 2007). In this way, recycled micronutrient reserves from thedeeper layers may maintain or even enhance their abundance in thetopsoil. In nature, Fe-oxides can be dissolved, especially in the presenceof organic acids and ligands that results from the root exudates and

ween Co and V in Lumbarda soils.

290 R. M. et al. / Catena 113 (2014) 281–291

breakdown of biologicalmaterial (Stipp et al., 2002), which is importantnot only for Fe supply but also for the supply of other essential traceelements.

5. Conclusions

Geostatistical analysis of the multivariate set of data revealed largespatial variability of soil properties within the study area of Lumbarda“polje”. Natural pedogeogenic signatures are being highly altered bylong term cultivation, but are still recognizable and two soil unitswere identified according to the World Reference Base (FAO/ISRIC/ISSS, 2006): SU1 Hypoluvic Arenosols, and SU2 Haplic Arenosols. Thevariation of soil properties with depth, especially sand and silt fractioncontent is likely related to coastal sedimentation processes.

In SU1, the reddish brown color originated fromFe oxide coatings onsand grains, whereas SU2 is characterized by the rise of pH with depthin the presence of calcite as a cementingmaterial. CEC was significantlyhigher in SU2. The variability of element content and distributionwithinthe soil profiles indicates different weathering stages of the sandylayers. In addition, diagrams of Cr/V and Co/V ratios indicate the differ-ent origin of sand deposits. Results suggest that soils of the study areaare developed from clastic sediments deposited by distinctive sourcesin different environments: eolian deposits in SU1, fluvial alluvia in SU2.

The differences in mineral distribution are likely caused by the na-ture of source area andmineral sorting by surface processes.Mineralog-ical and geochemical characterization of this grapevine growing site canbe used for further studies on “fingerprinting” of regional wines.

With regard to soil fertility and suitability for viticulture, the mostnotable characteristics are alkaline pH, low organic matter content,weak soil buffer capacity and low level of available nutrients, which in-dicate that fertility of soils is not sufficient to provide most of the nutri-ents required, except for potassium. Fertility conditions suitable forcultivation of high-quality grapes and water-holding capacity of sandysoils may be improved by organic matter enrichment. This study re-vealed the high influence of land use on soil characteristics, fertility,mineral nutrient acquisition from low-fertile sandy soil, and transloca-tion within the field.

Acknowledgments

This study was financed by the project “Improvement of vineyardmanagement of Vitis vinifera L. cv. Grk in the Lumbarda vineyard region(Croatia)” financed by the Neretvansko-Dubrovacka County and Munic-ipality of Lumbarda, as well as the project of theMinistry of Science, Ed-ucation and Sport of the Republic of Croatia, contract 178-1782221-2039“Spatial variability of potentially toxic metals in agricultural soils ofCroatia”.

References

Agostini, S., Chaturvedi, A., Kaplan Depolo, P., Tomasi, J., 2006. Role of agricultural & agro-pastoral practices in the formation of cultural landscapes: cultural determinants,commonalities and divergences in three continents. Cultural Landscapes in the 21.Century. Cultural Landscapes, Laws, Management, and Public Participation: Heritageas a Challenge of Citizenship Newcastle upon Tyne. University of Newcastle uponTyne.

Amorosi, A., Sammartino, I., 2007. Influence of sediment provenance on backgroundvalues of potentially toxic metals from near-surface sediments of Po coastal plain(Italy). Int. J. Earth Sci. 96, 389–396.

Bauluz, B., JoseMayayo,M., Fernandez-Nieto, C., Gonzalez Lopez, J.M., 2000. Geochemistryof Precambrian and Paleozoic siliciclastic rocks from the Iberian Range (NE Spain):implications for source-area weathering, sorting, provenance, and tectonic setting.Chem. Geol. 168, 135–150.

Bragato, G., Romic, M., Mosetti, D., Zovko, M., 2009. Soil variation in Grk vineyards of theKorcula Island, Croatia. In: Murisier, F. (Ed.), 32nd World Congress of Vine and Wine(Zagreb).

Brus, D.J., de Gruijter, J.J., van Groeningen, J.W., 2003. Designing spatial coverage samplesby the k-means clustering algorithm. Proceedings of the 8th International FZK/TNOConference on contaminated soil, Gent (Belgium), pp. 504–509.

Dane, J.H., Topp, G.C., 2002. Methods of Soil Analysis, Part 4 — Physical Methods. SSSABook Series, Vol. 5 (Madison).

Dawson, J.J.C., Smith, P., 2007. Carbon losses from soil and its consequences for land-usemanagement. Sci. Total. Environ. 382, 165–190.

Durn, G., Aljinovic, D., Crnjakovic, M., Lugovic, B., 2007. Heavy and light mineral fractionsindicate polygenesis of extensive terra rossa soils in Istria, Croatia. In: Mange, M.A.,Wright, D.T. (Eds.), Heavy Minerals in Use. Dev. Sedim, 58, pp. 701–737.

Egner, H., Riehm, H., Domingo, W.R., 1960. Untersuchungen uber die chemischeBodenanalyse als Grundlage fur die Beurteilung des Nahrstoffzustandes der Boden,II: Chemische Extractionsmetoden zu Phosphorund Kaliumbestimmung. KunglLantbrökshugskolans Ann., 26 199–215.

FAO/ISRIC/ISSS, 2006. World reference base for soil resources. World Soil Resources Re-ports 103.FAO, Rome.

Feng, Q., Endo, K.N., Cheng, G.D., 2002. Soil carbon in desertified land in relation to sitecharacteristics. Geoderma 106, 21–43.

Gilbert, M., Pammenter, N., Ripley, B., 2008. The growth responses of coastal dune speciesare determined by nutrient limitation and sand burial. Oecologia 156 (1), 169–178.

Goodall, D.W., 1966. A new similarity index based on probability. Biometrics 22, 882–907.Goodall, D.W., Ganis, P., Feoli, E., 1987. Probabilistic Methods in Classification: A Manual

for Seven Computer Programs. GEAD-EQ Report No. 7. Universita degli Studi diTrieste, Trieste.

Halamic, J., Miko, S., 2009. Geochemical Atlas of the Republic of Croatia. Croatian Geolog-ical Survey, Zagreb.

Hesp, P.A., Martinez, M.L., 2007. Disturbance processes and dynamics in coastal dunes. In:Johnson, E.A., Miyanishi, K. (Eds.), Plant Disturbance Ecology: The Process and the Re-sponse. Academic Press, Burlington, pp. 215–247.

Houba, V.J.G., Uittenbogaard, J., Pellen, P., 1996. Wageningen evaluating programmes foranalytical laboratories (WEPAL), organization and purpose. Commun. Soil Sci. PlantAnal. 27 (3–4), 421–431.

HRN ISO 11466, 2004. Soil Quality — Extraction of Trace Elements Soluble in Aqua Regia.International Organisation for Standardisation. Croatian Standard Institute.

HRN ISO 14235, 1998. Soil Quality — Determination of Organic Carbon by SulfochromicOxidation. International Organisation for Standardisation. Croatian Standard Institute.

Jurina, I., Ivanic, M., Vidovic, N., Mikac, N., Sondi, I., 2010. Mechanism of the land–seainteractions in the Neretva River delta (Croatia): the distribution pattern of sedimentsand trace elements. Rapports Commission Internationale pour l'explorationscientifique de la Mer Mediterranee 39 (Venice).

Kirkman, J.H., Basker, A., Surapaneni, A., MacGregor, A.N., 1994. Potassium in the soils ofNew Zealand — a review. J. Agric. Res. 37, 207–227.

Korolija, B., Borovic, I., Grimani, I., Marincic, S., Jagacic, T., Magas, N., Milanovic, M., 1977.Explanatory Text for General Geologic Map of Yugoslavia, 1: 100 000, Sheet Korcula,K 33–47. Institute of Geology, Zagreb.

Kovacevic Galovic, E., Ilijanic, N., Peh, Z., Miko, S., Hasan, O., 2012. Geochemical discrimi-nation of Early Palaeogene bauxites in Croatia. Geol. Croat. 65 (1), 53–65.

Lambeck, K., Yokoyama, Y., Purcell, T., 2002. Into and out of the Last Glacial Maximum:sea-level change during oxygen isotope stages 3 and 2. Quat. Sci. Rev. 21 (1–3),343–360.

Magliulo, P., Terribile, F., Colombo, C., Russo, F., 2006. A pedostratigraphic marker in thegeomorphological evolution of the Campanian Apennines (Southern Italy): thePaleosol of Eboli. Quat. Int. 156–157, 97–117.

Mange, M.A., Maurer, H.F.W., 1992. HeavyMinerals in Colour. Chapman and Hall, London.Martinez, M.L., Psuty, N.P., 2004. Coastal Dunes: Ecology and Conservation. Springer, New

York.McBride, M.B., 1994. Environmental Chemistry of Soil. Oxford University Press, New York.Mengel, K., Kirkby, E.A., 1987. Principles of Plant Nutrition, 2th Edition. International

Potash Institute, Worblaufen-Bern.Mikulcic Pavlakovic, S., Crnjakovic, M., Tibljas, D., Soufek, M.,Wacha, L., Frechen,M., Lackovic,

D., 2011. Mineralogical and geochemical characteristics of Quaternary sediments fromthe Island of Susak (Northern Adriatic, Croatia). Quat. Int. 234 (1–2), 32–49.

Milla, R., Castro-Diez, P., Maestro-Martinez, M., Montserrat-Marti, G., 2005. Environmen-tal constraints on phenology and internal nutrient cycling in the Mediterraneanwinter-deciduous shrub Amelanchier ovalis Medicus. J. Plant Biol. 7, 182–189.

Mpelasoka, B.S., Schachtman, D.P., Treeby, M.T., Thomas, M.R., 2003. A review of potassi-um nutrition in grapevines. Aust. J. Grape Wine Res. 9, 154–168.

Munsell Soil Color Charts, Munsell ColorRevised ed. GretagMacbeth, New Windsor, NewYork.

Novara, A., Pristina, L., La Mantia, T., Ruhl, J., 2011. Soil carbon dynamics duringsecondary succession in a semi-arid Mediterranean environment. Biogeosci.Discuss. 8, 11107–11138.

Pamic, J., Tomljenovic, B., Balen, D., 2002. Geodynamic and petrogenetic evolution ofAlpine ophiolites from the central and NWDinarides: an overview. Lithos 65, 113–142.

Papiernik, S.K., Lindstrom, M.J., Schumacher, T.E., Schumacher, J.A., Malo, D.D., Lobb, D.A.,2007. Characterization of soil profiles in a landscape affected by long-term tillage. SoilTillage Res. 93, 335–345.

Pavelic, D., Kovacic, M., Vlahovic, I., Wacha, L., 2011. Pleistocene calcareous aeolian–alluvial deposition in a steep relief karstic coastal belt (island of Hvar, easternAdriatic, Croatia). Sed. Geol. 239 (1–2), 64–79.

Podani, J., 2000. Introduction to the Exploration of Multivariate Biological Data. BackhuysPublishers, Leiden.

Reis, A.P., Menezes de Almeida, L., Ferreira da Silva, E., Sousa, A.J., Patinha, C., Fonseca, E.C.,2006. Assessing the geochemical inherent quality of natural soils in the Douro riverbasin for grapevine cultivation using data analysis and geostatistics. Geoderma 141(3–4), 370–383.

Rengel, Z., 2007. Cycling of micronutrients in terrestrial ecosystems. In: Marschner, P.,Rengel, Z. (Eds.), Nutrient Cycling in Terrestrial Ecosystems. Springer-Verlag, BerlinHeidelberg, pp. 93–113.

291R. M. et al. / Catena 113 (2014) 281–291

Robertson, A., Karamata, S., Saric, K., 2009. Overview of ophiolites and related units in theLate Palaeozoic–Early Cenozoic magmatic and tectonic development of Tethys in thenorthern part of the Balkan region. Lithos 108 (1–4), 1–36.

Rollinson, H.R., 1993. Using Geochemical Data: Evaluation, Presentation, Interpretation.Longman, Singapore.

Romic, M., Zovko, M., Romic, D., Bakic, H., 2012a. Improvement of vineyard managementof Vitis vinifera L. cv. Grk in the Lumbarda vineyard region (Croatia). Commun. SoilSci. Plant Anal. 43 (1–2), 209–218.

Romic, D., Romic, M., Zovko, M., Bakic, H., Ondrasek, G., 2012b. Trace metals in the coastalsoils developed from estuarine floodplain sediments in the Croatian Mediterraneanregion. Environ. Geochem. Health 34 (4), 399–416.

Soil Survey Division Staff, 1993. Agric. Handbk. 18. NRCS, Washington, DC.Stipp, S.L.S., Hansen, M., Kristensen, R., Hochella, M.F., Bennedsen, L., Dideriksen, K., Balic-

Zunic, T., Leonard, D., Mathieu, H.J., 2002. Behaviour of Fe-oxides relevant to contam-inant uptake in the environment. Chem. Geol. 190, 321–337.

Suric, M., Juracic, M., 2010. Late Pleistocene–Holocene environmental changes — recordsfrom submerged speleothems along the Eastern Adriatic coast (Croatia). Geol. Croat.63 (2), 155–169.

Suric, M., Richards, D.A., Hoffmann, D.L., Tibljas, D., Juracic, M., 2009. Sea-level change dur-ing MIS 5a based on submerged speleothems from the eastern Adriatic Sea (Croatia).Mar. Geol. 262 (1–4), 62–67.

Tripathi, J.K., Rajamani, V., 1999. Geochemistry of the loessic sediments on Delhi ridge, easternThar desert, Rajasthan: implications for exogenic processes. Chem. Geol. 155, 265–278.

Vital, H., Stattegger, K., 2000. Major and trace elements of stream sediments from the low-ermost Amazon River. Chem. Geol. 68 (1–2), 151–168.

White, R.E., 2003. Soils for Fine Wines. Oxford University Press, New York.Zdunic, G., Pejic, I., Karoglan Kontic, J., Vukicevic, D., Vokurka, A., Pezo, I., Maletic, E., 2008.

Comparison of genetic and morphological data for inferring similarity among nativeDalmatian (Croatia) grapevine cultivars (Vitis vinifera L.). J. Food Agric. Environ. 6(2), 333–336.