Bulk and clay mineral composition indicate origin of terra rossa soils in Western Herzegovina

Abstract

The B horizons of terra rossa soils developed on three different carbonate lithologies having variable insoluble residue contents were studied in Western Herzegovina. Comparison of their composition and properties illustrates to what extent the mineral, (especially clay mineral assemblage) and particle size distribution of those horizons and the insoluble residue of the underlying carbonate rocks can be used as indicators of the polygenetic nature of terra rossa in this region. Terra rossa B horizons have characteristic red colours, neutral to slightly acid pH, high base saturation with calcium as the predominant cation and high Chemical Index of Alteration (CIA). The CIA values obtained are generally in accordance with mineral composition and particle size distribution of the analysed B horizons. The predominant clay mineral phases in B horizons and related insoluble residues match. Kaolinite is the predominant clay mineral phase in the B horizons overlying carbonate rocks containing low amounts of insoluble residue, while smectite predominates in calcarenites areas with a high insoluble residue content.

Headnote

ABSTRACT

The B horizons of terra rossa soils developed on three different carbonate lithologies having variable insoluble residue contents were studied in Western Herzegovina. Comparison of their composition and properties illustrates to what extent the mineral, (especially clay mineral assemblage) and particle size distribution of those horizons and the insoluble residue of the underlying carbonate rocks can be used as indicators of the polygenetic nature of terra rossa in this region. Terra rossa B horizons have characteristic red colours, neutral to slightly acid pH, high base saturation with calcium as the predominant cation and high CIA (Chemical Index of Alteration). The CIA values obtained are generally in accordance with mineral composition and particle size distribution of the analysed B horizons. The predominant clay mineral phases in B horizons and related insoluble residues match. Kaolinite is the predominant clay mineral phase in the B horizons overlying carbonate rocks containing low amounts of insoluble residue, while smectite predominates in calcarenites areas with a high insoluble residue content. However, the presence of plagioclase, gibbsite, chlorite-venniculite mixed layer mineral and Vermiculite in B horizons overlying carbonate rocks containing low amounts of insoluble residue support a polygenetic origin for the terra rossa. In contrast, terra rossa formed on calcarenites containing high amounts of insoluble residue might have formed almost exclusively from the parent carbonate rock although some influence of external materials (e.g. gibbsite) cannot be excluded. This investigation shows that in Western Herzegovina, an area with no important aeolian input, the content and mineral composition of carbonate rock insoluble residue plays a major role in terra rossa composition. We can tentatively conclude that the lower the insoluble residue content of the parent materials, the greater is the expectation of a more polygenetic origin for the terra rossa.

Keywords: terra rossa, B horizons, parent materials, (clay) mineralogy, Western Herzegovina

1. INTRODUCTION

Terra rossa is a reddish clayey to silty/clayey soil developed over limestones and dolomites and is especially widespread in the Mediterranean region. A diagnostic feature of terra rossa is a bright red colour originating from rubification, a pedogenic process in winch preferential fonnation of haematite over goethite takes place. Due to the underlying highly penneable carbonate rocks, terra rossa is well aggregated and drained, lias a slightly alkaline to neutral pH and abase complex almost completely saturated by calcimn and/ or magnesium. Terra rossa is fonned as a result of (1) décalcification, (2) rubification and (3) bisiallitisation and/or monosiallitisation (DURN et al., 1999). Terra rossa is classified as Alfisol (Haploxeralf or Rhodoxeralf), Ultisol, Inceptisol (Xerochrept) and Mollisol (Argixeroll or Haploxeroll) in Soil Taxonomy (SOIL SURVEY STAFF, 1975). According to the FAO system (FAO, 1974) terra rossa is recognised as Luvisol (Chromic Luvisol), Phaeozem (Haplic Phaeozem or Luvic Phaeozem) and Cambisol. In other classification systems using the Mediterranean climate as the major soil differentiating criterion, the tenu terra rossa is used to describe the soil subclass "Modal Fersiallitic Red soil" when situated on limestones (DUCHAUFOUR, 1982). Several national soil classifications (e.g. Croatia, Italy, Israel) retain the tenu "terra rossa" for limestone-derived red soils. The Croatian classification puts terra rossa in the class of Cambie soils (SKORIC, 1985). Although different authors have considered terra rossa to be a soil, vetusol, relict soil (non-buried-palaeosol), palaeosol or pedosedimentary complex, most scientists consider terra rossa a polygenetic relict soil, formed during the Tertiary and/or hot and humid periods of the Quaternary (e.g. ALTAY, 1997; BRONGER & BRUHN-LOBIN, 1997; DURN et al" 1999).

The nature and relationship of terra rossa to the underlying carbonates is a long-standing problem which lias resulted in different opinions with respect to the parent material and origin of terra rossa. The most widely accepted theory is that terra rossa has developed from the insoluble residue of carbonate rocks (TUCAN, 1912; KISPATIC, 1912; KUBIËNA, 1953; MARIC, 1964; CIRIC & ALEKSANDROVIC, 1959; PLASTER & SHERWOOD, 1971; SKORIC, 1979, 1987; BRONGER et al" 1983; MORESI & MONGELLI, 1988). Other authors have emphasised that the addition of various external materials (e.g. afigeolian dust, volcanic debris, clastic sedimentary particles, bauxite particles) might have masked the influence of limestone and dolomite residues as the primary parent material of terra rossa (BALAGH & RUNGE, 1970; YAALON & GANOR, 1973; SINKOVEC, 1974; OLSON et al" 1980; MACLEOD, 1980; JACKSON et ah, 1982; DANIN et ah, 1983; RAPP, 1984; JAHN et ah, 1991; NIHLEN & OLSSON, 1995; ALTAY, 1997; DURN et ah, 1999, DURN, 2003; DURN et ah, 2007; MUHS et. ah, 2012). DURN et ah (2007) concluded that in some isolated karst terrains, terra rossa may have formed exclusively from the insoluble residue of limestone and dolomite, but it most commonly comprises a variety of external materials that were carried to the carbonate terrain by various transport mechanisms. Although it is difficult to estimate accurately the proportions of external materials embedded in the terra rossa soils, DURN et ah (2007) estimated that in such a nonisolated karst terrain (such as Istria) their contribution might have been up to 50%. MUHS et ah (2012) investigated the origin of terra rossa soils overlying very pure carbonate substrates of Quaternary age in Bermuda. Based on the detailed geochemical analyses of trace elements that are immobile in the soil-forming enviromnent, they discovered that terra rossa soils have been influenced by a combination of LRT dust from Africa and local volcanic bedrock. They also concluded that soils on islands in a very broad latitudinal belt of the western Atlantic margin have been influenced by African LRT dust inputs over much of the past 500 ka.

BOERO & SCHWERTMANN (1989) suggested that terra rossa is formed in a specific pedo-enviromnent which is characterised by an association of Mediterranean climate, high internal drainage due to the karstic nature of a hard limestone and neutral pH conditions. They also concluded that it is of little relevance for the process of rubification whether the primary Fe sources are autochthonous or allochthonous as long as the general pedoenviromnent remains essentially suitable for the formation of terra rossa. Based on new field and petrographic evidence, MERINO & B ANERJEE (2008) proposed a new theory of terra rossa formation, by replacement of limestone by authigenic clay at a narrow reaction front and explained why terra rossa and karst are associated. B ANERJEE & MERINO (2011) successfully model the new terra rossa-fonning process quantitatively and compared model-calculated rates of formation of terra rossa to rates obtained palaeomagnetically.

The tenu reddish soils refers to all soils of different shades of red colour developed on carbonate sediments. Reddish soils include various types of soils in different phases of soil development: from youngest reddish or brownish calcomelanosols, through terra rossa and calcocambisols to luvisols, which are genetically considered the oldest and most developed soils on carbonate sediments. The basic criterion for distinguishing reddish soils from other soils is the red colour of the diagnostic horizon. Chemical and physical properties of soils in Western Herzegovina, especially reddish soils were subject to investigation by several researchers (BUKOVAC, 1950; RESULOVIC et ai, 1963; KURTOVIC, 1971; 1979). B ABIC (1989) concluded that the main clay minerals in terra rossa developed on flysch in descending order are smectite, kaolinite and micaceous minerals respectively. TVICA (2008) found high base saturation in terra rossa soils from Herzegovina with calcimn being the predominant cation. Terra rossa is the most frequent soil type in Western Herzegovina (Bosnia and Herzegovina). Among 111.944 ha of reddish soils, terra rossa covers 44.554 ha (39,80%) (Fig. 1). According to KURTOVIC (1973; 1979), terra rossa has developed on different types of carbonate rocks (e.g. rudist limestone, brecciated limestone, calcarenite). Therefore, terra rossa soils in Western Herzegovina represent promising material to investigate the influence of the underlying carbonates on the mineral composition and grain-size distribution of terra rossa.

The physical, chemical and mineralogical features of the B horizons of terra rossa soils developed on three different carbonate lithologies have been investigated here. The aim of this research was to compare the mineral composition of bulk samples and clay fraction of terra rossa B horizons with the insoluble residues of the underlying carbonate lithologies. This should also show the influence of the insoluble residue content and its grain size on the B horizons of terra rossa for different carbonate lithologies. Finally, this work should show to what extent the mineral, (especially clay mineral assemblage) and particle size distribution of the B horizons and insoluble residue of the underlying carbonate rocks can be used as indicators of the polygenetic nature of terra rossa in Western Herzegovina.

2. MATERIALS AND METHODS

2.1. Study area

The research area is located in the Western Herzegovina region of Bosnia and Herzegovina. This region completely lies within the Dinarides mountains, which are mostly composed of Mesozoic platform carbonates deposited on the Adriatic Carbonate Platform (AdCP) (for details see: DRAGICEVIC & VELIC, 2002; TISLJAR et al., 2002; VLAHOVIC et al., 2005). The area between Citluk and Siroki Brijeg is formed from Cretaceous carbonate rocks (various types of limesto- nes and dolomites), Tertiary carbonates and elastics and Qua- ternary deposits (MOJICEVIC & LAUSEVIC, 1971; RAIC et al, 1975). According to summarized data from MOJI- CEVIC & LAUSEVIC (1973) and RAIC & PAPES (1977), the following lithostratigraphic units can be observed in the field: Lower Cretaceous (Berriasian to Aptian) carbonates; Albian to Cenomanian dolomites and limestones; Cenoma- nian to Turonian chondrodont and rudist limestones; Turo- nian to Coniacian rudist limestones; Palaeocene (Liburnian) limestones; Palaeocene to Lower Eocene foraminiferal lime- stones; Middle Eocene elastics; Neogene (Miocene) elastics and carbonates; and Quaternary deposits (Figure 2). Several tectonic phases resulted in folded and overthrust relationships of the Cretaceous and Tertiary units in the field, together forming the Stolac-Citluk tectonic unit (RAIC & PAPES, 1977). Significant amounts of bauxite deposits accumulated during the continental phases (at the boundaries between the Cretaceous and Tertiary and in the Middle Eocene) in the synforms of the developed palaeorelief. In the Neogene, clas- tic and lacustrine carbonate sedimentation predominated. Quaternary deposits accumulated significantly in the low- lands (in karstic dolinas and poljes and in the river valleys) as fluvioglacial, limnoglacial, alluvial, deluvial, proluvial and colluvial material. Most of the karstic lowlands in the re- search area are filled with reddish soils, especially terra rossa, overlying different lithostratigraphical units.

2.2. Investigated locations

Three terra rossa profiles at three different locations (profile 2 at Kocerin, profile 5 at Citluk and profile 9 at Uzariéi) were opened in Western Herzegovina (Figs. 2 and 3). At each lo- cation GPS coordinates were recorded using Mobile Mapper 6, Magellan professional, as follows: X=6,456,271m, Y=4,805,093m (Kocerin); X=6,474,991m, Y=4,786,373m (Citluk); X=6,469,858m, Y=4,801,696m (Uzariéi). At Koce- rin medium deep terra rossa is developed on brecciated mi- critic limestone (Fig. 3A). At Citluk Luvic terra rossa over- lies calcarenite (Fig. 3B). Shallow terra rossa in Uzariéi formed on top of bioclastic floatstone (Fig. 3C). At Kocerin and Uzariéi the (B)rz horizon of terra rossa was sampled while at Citluk samples were taken from the Bt horizon. (B) rz horizons according to SKORIC et al. (1985) can be cor- related with Bw horizons according to FAO (2006). In our investigation this horizon is considered to be in direct con- tact with the carbonate rock. The lowermost Bt horizon at Citluk site is also developed on carbonate rock. However, since this profile is that of a Luvic terra rossa, (B)rz (ie. : a rendzic B horizon) horizon was not present and the Bt hori- zon was sampled accordingly.

2.3. Field and laboratory methods

Soil pits were dug to the contact with carbonate rock, while soil characterization and sampling were done in accordance with SKORIC (1982; 1986). Samples of carbonate rocks were collected immediately from below the soil profiles. Disturbed soil samples were collected from all genetic soil horizons and air-dried afterwards. A portion of each dis- turbed soil sample was gently crushed and sieved through a 2 imn sieve for physical, chemical and mineralogical anal- yses. Soil particle size distribution was determined by the pipette method with sieving and sedimentation after disper- sion with sodium pyrophosphate, and interpreted according to FAO (2006). The <2 pm fraction of soil samples was separated by sedimentation in a cylinder and quantitatively obtained after the appropriate settling time. Soil pH in H20 and in 1 M KCl was measured in 1:2.5 suspension, while the cation exchange capacity (CEC) and base saturation (BS) level were determined using barium chloride accord- ing to HRN ISO 11260 (2004). The chemical composition of soil samples was determined by the coimnercial ACME Analytical Laboratory, Canada. Major oxides were deter- mined by X-ray fluorescence (XRF) spectrometer following LiB02 fusion.

Thin sections of carbonate rocks were analysed using a Leica DM/LSP petrographic microscope with plane-polari- zed (ppl) and crossed-polarized light (xpl).

Samples of carbonate rocks were carefully crushed and sieved to pass through a 2 and 4 mm sieve. Fragments in the 2-4 imn range were carefully cleaned in water and only those without impurities were picked. To remove carbonates, the picked fragments of carbonate rocks (2-4 mm) were treated with a 1 M NaOAc solution buffered at pH 5 with HO Ac (JACKSON, 1979; TASSIER et al" 1979.) The particle size analysis of the insoluble residues was determined after dis- persion in water and ultrasonic treatment. Fractions >63 pm were obtained by wet sieving. The <2 pm fraction was sep- arated by sedimentation in cylinder and quantitatively ob- tained after the appropriate settling time. The remaining cyl- inder content was calculated as representing the 2-63 pm fraction.

The mineral composition of <2 imn and <2 pm fractions of soils and the insoluble residue of carbonate rocks (bulk insoluble residue and <2 pm fraction of insoluble residue) was determined by X-ray powder diffraction (XRD) using a Philips diffractometer (graphite monochromator, CuKp ra- diation, proportional counter). XRD patterns of clay fractions were made after the following treatments: (a) air-drying, (b) glycerol solvation, (c) glycol solvation, (d) heating to 550°C for 2 h and (e) dissolution in HCl ( 18%) for 24 h. The DMSO- treatment was used to differentiate kaolinites which fonn in- tercalation compounds with DMSO from kaolinites which do not intercalate withDMSO (RANGE et al.,1969). By means of semi-quantitative XRD analysis, the amounts of quartz, plagioclase, K-feldspar, haematite, goethite, anatase and gibb- site were detennined in the <2 mm and <2 pm fractions of soil samples, bulk insoluble residues and <2 pm fraction of insoluble residues. An external standard was applied, by measuring the relative intensities of characteristic diffraction lines. The identification of clay minerals was generally based on the methods outlined by BROWN (1961), BRINDLEY & BROWN (1980) and MOORE & REYNOLDS (1989). The tenu "illitic material" was used as defined by SRODOÑ (1984) and SRODOÑ & EBERL (1984). The tenu "MC" was used for mixed-layer clay minerals in which the type of in- terstratification and constituent clay minerals were not readily identifiable. Semi-quantitative estimates of clay minerals in the <2 pm fraction were based on the relative intensities of characteristic X-ray peaks following the method of JOHNS et al. (1954). Estimated quantities of minerals were presented with Xs, but no quantitative value was assigned to each X.

3. RESULTS

3.1. Carbonate rocks

Results obtained from the micropetrographic analysis of the underlying lithologies at Kocerin, Citluk and Uzarici showed good correlation with previously determined litho stratigra- phy presented on the geological map of the investigated area (Fig. 2). The observed profiles with terra rossa were deve- loped on different litho stratigraphic units as follows: (i) Al- bianto Cenomanian dolomites and limestones (Kocerin); (ii) Middle Eocene elastics (Citluk) and (iii) Cenomanian to Turanian chondrodont and rudist limestones (Uzarici). At Kocerin, brecciated micritic limestone, showing recrystal- lisation and development of meteoric diagenetic features was recognized. It is interpreted as a diagenetic carbonate crust, developed in the Albian to Cenomanian dolomite and limestone lithostratigraphic unit (Fig. 4). At Citluk, a calca- renite was identified containing predominantly well-sorted, intra-basinal carbonate material (bioclasts and intraclasts), accompanied together with a minor proportion of siliciclas- tic material (mainly quartz grains), cemented by calcitic ce- ment (Fig. 5). Rudist and chondrodont limestone, identified as the bioclastic floatstone (DUNHAM, 1962) is observed at Uzarici. It contains large (up to several centimetres) bio- clasts of chondrodonts and rudists, accompanied by some benthic forains, together embedded in a micritic matrix (Fig. 6). This bioclastic limestone belongs to the Cenomanian to Turanian chondrodont and rudist limestones lithostrati- graphic unit. According to the observed petrographic char- acteristics and bioclast assemblage, it belongs to the Middle Eocene elastics lithostratigraphic unit.

3.2. Soil chemical properties

Analysed B horizons of terra rossa have neutral to slightly acid pH and high base saturation with calcium as the pre- dominant cation, followed by magnesium (Table 1). The highest pH (H20) was observed in the (B)rz horizon devel- oped on the bioclastic floatstone (Uzarici) while the Bt hori- zon in the Luvic terra rossa overlying calcarenite lias the low- est pH (H20). As for pH, the BS and humus content were lowest in the Bt horizon at Citluk and highest in the (B)rz ho- rizon at Uzarici. As expected, the lowest CEC value was ob- served in the Bt horizon at Citluk where the lowest pH (H20) and humus content was detected. Namely, the Luvic terra rossa at this location is considered the most developed soil among the three analysed locations, and, in addition contains the lowest amount of the clay fraction (Table 2). It is also in accordance with WRIGHT & FOSS (1972), who concluded that CEC in the lower horizons of soil profiles depended on the clay content and the mineral composition of clay fraction.

3.2. Particle size analysis

The B horizons of terra rossa profiles at Kocerin, Citluk and Uzarici have characteristic red colours and are com- posed of clay, clay loam and clay respectively (Table 2). The insoluble residue content of the underlying carbonate rocks is very variable (Table 3). It is extremely low in Uzarici (0.09 wt.%), low in Kocerin (0.61 wt.%) and high in Citluk (18.55 wt.%). All B horizons are enriched in the clay fraction and depleted in the silt fraction compared to the insoluble residue (Tables 2 and 3). The content of the sand fraction in the B horizon from Kocerin and Uzarici is higher compared to the sand content in the insoluble resi- due of the underlying carbonate rocks (Tables 2 and 3). In contrast, the Bt horizon developed on calcarenite (Citluk), which has the highest insoluble residue content dominated by the sand fraction is depleted in the sand fraction com- pared to the insoluble residue.

3.4. Geochemical indicators of weathering

In order to characterize the investigated B horizons geo- chemically, we used the Chemical Index of Alteration (CIA) proposed by NESBITT & YOUNG (1982) and considered the measure of feldspar minerals weathering and their hydra- tion to fonn clay minerals (SHELDON & TABOR, 2009) and Ti/Al molecular ratio, which can also be used as prov- enance indicator (e.g. SHELDON, 2006; STILES & STENS- VOLD, 2008). CIA values range from 84.73 in Citluk to 93.32 in Kocerin and can be considered as very high (Table 4). The Ti/Al ratios in bulk samples range from 0.040 in Kocerin to 0.060 in Citluk (Table 4).

3.5. Bulk and clay mineralogy

Mineralogical analyses were perfonned on the <2 mm and <2 pm fractions of the B horizons (Tables 5 and 6; Figs. 7 and 9) and the insoluble residue of carbonate rocks (bulk in- soluble residue and <2 pm fraction of insoluble residue) (Ta- bles 7 and 8: Figs. 8 and 10). Though the mineralogical com- position of non-clay minerals in all three B horizons was relatively uniform (Table 5), significant differences in clay mineral composition among the investigated B horizons were observed (Table 6). Each B horizon contained quartz, K-feldspar, haematite, goethite, gibbsite, kaolinite, MC and XRD-amorphous inorganic compounds. Both kaolinite which does not fonn intercalation compounds with DMSO and kaolinite which intercalates with DMSO were detected. The presence and distribution of plagioclase, anatase, smec- tite, venniculite, micaceous clay minerals (illitic material and mica) and chlorite-venniculite mixed layer mineral (C/V) varied among the different B horizons. The Bt horizon from Citluk contains the highest amount of quartz (in both fractions) and only in this sample was the presence of pla- gioclase confirmed with certainty. Although gibbsite was recognized in all B horizons, its content is highest in the (B) rz horizon from Kocerin (in both fractions).

Kaolinite is the predominant clay mineral phase in the clay fraction of the (B)rz horizons from Kocerin and Uzarici while smectite is the predominant clay mineral phase in the clay fraction of the Bt horizon from Citluk (Table 6; Figs. 7 and 9). Kaolinite is followed by venniculite and C/V as the main mineral phases in Kocerin and by C/V and illitic mate- rial in Uzarici. It is important to emphasize that illitic mate- rial was confinned with certainty only in the (B)rz horizon from Uzarici. Kaolinite and MC follow smectite in Citluk in tenus of quantity.

Significant differences in both the non-clay and clay mineral composition of the analysed insoluble residues (IR) were observed (Tables 7 and 8). Each IR contained quartz, goethite, kaolinite and an XRD-amorphous inorganic com- pound. Both kaolinite which does not fonn intercalation compounds with DMSO and kaolinite which intercalates with DMSO were detected. The presence and distribution of plagioclase, K-feldspar, haematite, anatase, gibbsite, smec- tite, micaceous clay minerals (illitic material and mica), MC, C/V and illite/smectite mixed-layer mineral (I/S) varied among the different IR. IR from Citluk contains the highest amount of quartz (in both fractions), and plagioclase andK- feldspar were detected in this sample alone. IR from Kocerin contains the highest amount of goethite and is the only sam- ple with gibbsite (in both fractions).

Kaolinite is the predominant clay mineral phase in the clay fraction of IR from Kocerin and Uzarici while smectite is the predominant clay mineral phase in the clay fraction of IR from Citluk (Table 8; Figs. 8 and 10). Kaolinite is followed by smectite, illitic material and ES in Kocerin and is the only clay mineral phase in Uzarici. Smectite is followed by illitic material, kaolinite and I/S in Citluk.

4. DISCUSSION AND CONCLUSION

The terra rossa soils analysed from Western Herzegovina oc- cur over carbonate rocks of different ages and lithologies (Figs. 2, 4, 5 and 6) having very variable IR contents (Table 3) ranging from extremely low (0.09 wt.% in Uzarici) to high (18.55 wt.% in Citluk) values. Terra rossa B horizons have characteristic red colours (Table 1), neutral to slightly acid pH and high base saturation, with calcium as the pre- dominant cation, followed by magnesium (Table 1). It is in accordance with TVIC A (2008) who observed high base sat- uration in terra rossa soils from Herzegovina with calcium as the predominant cation.

Very high CIA values in the B horizons of the analysed soils clearly indicate intensive weathering (Table 4). Accord- ing to SHELDON & TABOR (2009), parent materials that have already been cycled as sediments or which are clay- rich, may start out with CIA values of 60 to 70%. They state that as weathering progresses from, for example, microcline to illite and kaolinite, CIA values would increase from 50 to 75 (pure illite) and 100 (pure kaolinite), respectively. The CIA values obtained in this study are generally in accord- ance with the mineral composition and particle size distribu- tion of the analysed B horizons. Traces of K-feldspar and plagioclase, the predominance of kaolinite as well as the clay fraction content are in favour of high CIA in the (B)rz hori- zons from Kocerin and Uzarici (Tables 2, 4, 5 and 6). The slightly lower CIA value observed in the Bt horizon from Citluk is due to the significantly lower clay fraction (higher amount of silt and especially sand fraction) and to smectite as the predominant clay mineral phase. However, it has to be stressed that CIA values are partly masked by the pres- ence of gibbsite (i.e. aluminium hydroxide) that was detected in all the analysed B horizons. The effect of particle size dis- tribution is also evident in the Ti/Al ratios because the high- est ratio was observed in the Bt horizon from Citluk with the highest amount of sand fraction and, likewise, the highest Ti content, while the lowest ratio was found in the (B)rz hori- zon from Kocerin with the highest clay content. Unfortu- nately, due to this effect of particle size, the Ti/Al ratio could not be used as a provenance indicator.

The content of the insoluble residue may indicate that in the case of Kocerin and Uzarici locations, an excessive thickness of carbonate rocks must have been dissolved to fonn the (B)rz horizons of terra rossa, and that also the ex- tent of the preservation of that residue must have been very high. In contrast, it is quite plausible that the Bt horizon at Citluk might have fonned from the insoluble residue of the calcarenite. Although the insoluble residue of carbonate con- tent of carbonate rocks below the (B)rz horizons in Kocerin and Uzarici is not compatible with the development of tena rossa entirely by the dissolution of carbonate rocks, its grain size distribution provides a different picture. Namely, if terra rossa has developed only from the insoluble residue of lime- stone or dolomite, its clay content, due to weathering should be higher than that observed (Tables 2 and 3). In order to re- solve such questions, the clay fraction may be of great im- portance because its composition could be a result of differ- ent processes which may have taken place during soil fonnation. Namely, soils may contain "detritaf'clays inheri- ted from the parent material (in this case soil clays represent the IR of carbonate rock) and they may contain clays fonned by sequential weathering of unstable parent minerals of the IR (including clay mineral phases). There are also neofor- med (pedogenic) clay minerals and clays added to the soil due to allochthonous inputs (e.g. aeolian dust, volcanic de- bris and clastic sedimentary particles; DURN et al., 2007). This is even more complex when soils are fonned on car- bonate rocks that are extremely low in non-carbonate min- eral phases.

With regard to the previous statement, the situation is clearest at Citluk because there the calcarenite contains a high amount of IR, 18.55 wt.% respectively (Table 3). Luvic terra rossa at this location is thickest (Table 1) and is consid- ered to be the most developed soil of the three analysed lo- cations. The clay mineral composition in the Bt horizon and IR match quite well (Tables 6 and 8). The presence of smec- tite as the predominant clay mineral phase in both the IR and Bt horizons clearly indicates that this mineral phase was in- herited from the parent material. Kaolinite is present in both the IR and Bt horizons and can also be, at least partly, con- sidered as inherited. Namely, the Bt horizon contain both types of kaolinite while kaolinite which forms intercalation compounds with DMSO is the predominant type of this min- eral phase in the IR. This may indicate that the kaolinite which does not intercalate with DMSO is predominantly pedogenic kaolinite, i.e., anauthigenic mineral in terra rossa, while kaolinite which intercalates with DMSO is inherited from the kaolinite containing parent material which means it is of a lithogenic origin (DURN et al., 1999). Illitic mate- rial and I/S detected in IR were not found with certainity in the Bt horizon probably due to their very low content. How- ever, the presence of MC in the Bt horizonmay indicate these phases were the source material for more weathered clay minerals. A much higher content of kaolinite would be ex- pected in the Luvic terra rossa but smectite seems to be the (meta)stable mineral phase in this pedoenviromnent because the pH was not low enough to inhibit smectite stability, (e.g. DOUGLAS, 1982; KARATHANASIS & HAJEK, 1984 ) and Ca and Mg are the predominant cations in the soil solu- tion (Table 1). Due to their chemical composition, the dis- solution of smectite minerals is driven by different mecha- nisms. In particular, hydrolysis of octahedral Mg is interpreted as the primary driving force behind the dissolu- tion of saponite in dilute acid, while reduction should play a major role in the decomposition of nontronite (RYAN et al., 2008). A higher content of quartz and much lower con- tents of unstable mineral phases including plagioclase and K-feldspar in the Bt horizon compared to IR (Tables 5 and 7) also favour calcarenite as the parent material for the Lu- vic terra rossa at Citluk. Therefore we can conclude that both the mineral composition and the particle size distribution of the Bt horizon and IR suggest calcarenite as terra rossa par- ent material. Minor external material contributions by vari- ous transport mechanisms cannot be excluded as docmnented by gibbsite which we consider to be the only external mate- rial in terra rossa at this location.

As previously stated, in the case of Kocerin and Uzarici, the IR content of carbonate rocks below terra rossa may in- dicate that an excessive thickness of carbonate rocks was dissolved to form the (B)rz horizons, and that the degree of the preservation of that residue must have been very high. Kaolinite is the predominant clay mineral phase in both the IR and the Bt horizon at those locations. We tentatively pro- pose the same origin of kaolinite as in the case of Citluk. However, two clay mineral phases (C/V and venniculite) were detected in the (B)rz horizons from these locations that were not found in the IR of brecciated micritic limestone (Kocerin) and bioclastic floatstone (Uzarici) (Tables 6 and 8).

We tentatively propose three possible explanations for the C/V and venniculite origin in the analysed (B)rz horizons with the first or/and third one being the most probable. Firstly, C/V and venniculite are considered as soil clay min- erals because those mineral phases were not detected in the IR of the conesponding carbonate rocks. The presence of C/V may indicate that both mineral phases fonned as a re- sult of chlorite destabilization, with C/V as an intennediate step during the venniculitization process of chlorite sensu WILSON (2004). This explanation is valid only when chlo- rite was present as an allochthonous mineral phase because this mineral was not detected in the IR of corresponding car- bonate rocks. The presence of venniculite in (B)rz horizon from Kocerin and lack of this clay mineral in (B)rz horizon from Uzarici can probably be attributed to the lower pH of Kocerin (Table 1).

Secondly, it is possible that venniculite formed by weathering of the illite, mineral phase detected in the IR of carbonate rocks from Kocerin and Citluk. For example, OT- TNER et al. (2013) state that illite can be the source material for more weathered clay minerals in the Oberlab loess-pal- aeosol sequence in Upper Austria. However, the lack of illite in the IR from Uzarici, as well as the lack of venniculite in the Bt horizon from Citluk do not support this explanation. This second explanation also does not provide an answer for the origin of C/V

Thirdly, it is possible that both mineral phases were not fonned in soil from allochthonous chlorite but are derived as mineral phases already present in "external" material. The presence of plagioclase as a trace mineral phase in Kocerin (plagioclase was not detected in IR of brecciated micritic limestone) and the presence of gibbsite in both (B)rz hori- zons locations support both the first and/or the third expla- nation.

We can conclude that both the mineral composition of the (B)rz horizon and IR as well as the very low IR content of carbonate rocks below terra rossa strongly suggest a sub- stantial contribution of external material during the genesis of terra rossa in Kocerin and Uzarici compared to Citluk. The positions of the three investigated locations on the geo- logical map (Fig. 2) clearly indicate that the sampling sites in Kocerin and Uzarici are situated in the vicinity of Quater- nary deposits which could have been the possible source of external materials while Citluk is situated in a more isolated position where the influence of external materials is less im- portant. Although the IR of brecciated micritic limestone from Kocerin contains gibbsite, (probably from cracks in limestones and, therefore, of secondary origin), we relate the presence of this mineral phase in all B horizons to reworked particles of Palaeogene bauxites which are sporadically present in the area of investigation (RAIC & PAPES, 1977).

Based on their studies of terra rossa soils in Istria DURN et al. (2007) concluded that in some isolated karst terrains, terra rossa may have fonned exclusively from the IR of lime- stone and dolomite, but it is most commonly composed of a variety of external materials, including aeoliandust, volcanic debris and clastic sedimentary particles that were carried to the carbonate terrain by various transport mechanisms. They found that the most likely additional flux influencing terra rossa formation in Istria is aeolian dust, followed by flysch sediments and their contribution might have been up to 50%. Based on investigation of the B horizons of terra rossa on three locations in Western Herzegovina we can conclude the following:

(1) terra rossa situated on isolated carbonate rocks contain- ing high amount of IR may have fonned almost exclu- sively from the parent carbonate rock (Citluk) although some influence of external materials cannot be excluded (e.g. gibbsite).

(2) tena rossa fonned on non-isolated carbonate rocks con- taining low amounts of IR (Kocerin and Uzarici) clearly shows the influence of external materials on its genesis (e.g. chlorite, plagioclase, gibbsite) and can be regarded as polygenetic soil (DURN et al., 1999; 2007). It is im- portant to stress that compared to Istria where loess dep- osition lias been a recunent process since the early Mid- dle Pleistocene and influenced tena rossa formation (DURN et al., 1999; 2007) no data on loess deposition in Western Herzegovina is reported in the literature.

ACKNOWLEDGMENT

This research was supported by Croatian Science Fondation (project 2504).

References

REFERENCES

ALTAY. I. (1997): Red Mediterranean soils in some karstic regions of Taurus mountains, Turkey - Catena, 28, 247-260.

BABIC, S. ( 1989): Physical, chemical and mineralogical characteristics of terra rossa and eutric red-brown (eutric cambisol) of Herzegovi- na, Zemljiste i biljka, 38, 7-17, Beograd.

BALAGH. T.M. & RUNGE. E.C.A. (1970): Clay-rich horizons over limestone, illuvial or residual.- Soil Sei. Soc. Am. Proc., 34, 534-536.

BANERJEE, A. & MERINO, E. (2011 ): Terra rossa genesis by replace- ment of limestone by kaolinite. III. Dynamic quantitative model. The Journal of Geology. 119. 259-274.

BOERO. V. & SCHWERTMANN. U. (1989): Iron oxide mineralogy of terra rossa and its genetic implications - Geoderma, 44, 319-327.

BRINDLEY. G.W. & BROWN. G. (1980): Crystal structures of clay minerals and their X-ray identification. Miner. Soc., London, 495 p.

BRONGER. A. & BRUHN-LOBIN, N. (1997): Paleopedology of Ter- rae rossae - Rhodoxeralfs from Quaternary calcarenites in NW Mo- rocco- Catena, 28, 279-295.

BRONGER. A.. ENSLING. J.. GÜTLICH. P. & SPIERING. H. (1983): Rubification of terrae rossae in Slovakia: a Mösbauer effect study. - Clays Clay Miner.. 31. 269-276.

BROWN, G. (1961): The X-ray identification and crystal structures of clay minerals.- Miner. Soc., London, 544 p.

BUKOVAC, P. (1950): Prilog poznavanjutlazapadne Hercegovine, Pol- joprivredna znanstvena smotra, sv. 12, Zagreb.

CIRIC. M. & ALEKSANDROVIC. D. (1959): A view on the genesis of terra rossa - Zbomik radova Poljoprivrednog fakulteta, 7, 1-12, Beograd.

DANIN. A.. GERSON. R. & CARTY. J. (1983): Weathering patterns on hard limestone and dolomite by endolithic lichens and cyanobacte- ria: supporting evidence for eolian contribution to terra rossa soil.- Soil Science. 136.213-217.

DOUGLAS. L.A. (1982): Smectites in acidic soils.-In: VAN OLPHEN. H. & VENIALE, F. (eds.): Proc. Int. Clay Conf. in Bologna, Pavia. Elsevier, Amsterdam, 635-640.

DRAGICEVIC. I. & VELIC. I. (2002): The Northeastern Margin of the Adriatic Carbonate Platform-Geol. Croat, 55/2, 185-232, Zagreb.

DUCHAUFOUR, P. (1982): Pedology: Pedogenesis and Classification- Allen and Unwin, London, 448 p.

DUNHAM, J.B. (1962): Classification of carbonate rocks according to depositional texture.- In: HAM, W.E. (ed.): Classification of car- bonate rocks. Amer. Assoc. Petrol. Geol. Mem., 1, 108-121.

DURN, G. (2003 ): Terra rossa in the Mediterranean region: parent ma- terials, composition and origin.-Geol. Croat, 56/1, 83-100, Zagreb.

DURN. G.. OTTNER. F. & SLOVENEC. D. (1999): Mineralogical and geochemical indicators of the polygenetic nature of terra rossa in Istria, Croatia-Geoderma, 91, 125-150.

DURN. G.. ALJINOVIC. D.. CRNJAKOVIC. M. & LUGOVIC. B. (2007): Heavy and light mineral fractions indicate polygenesis of extensive terra rossa soils in Istria Croatia.- In: MANGE, M.A & WRIGHT, D.T. (eds.): Heavy Minerals in Use 701-737, Develop- ments in Sedimentology, 58, Elsevier.

FAO (1974): Soil Map of the World. 1:5 Mill.. Volume 1. Legend.- Unesco, Paris.

FAO (2006): Guidelines for Soil Description, 4th edition. FAO, Rome.

HRN ISO 11260 (2004): Soil Quality - Determination of Effective Cat- ion Exchange Capacity and Base Saturation Level Using Barium Chloride Solution (ISO 11260:1994+Cor 1:1996).

JACKSON, M.L. (1979): Soil chemical analysis-advanced course. Soil Science Dept., University of Wisconsin, Madison, 250 p.

JACKSON. M.L., CLAYTON. R.N.. VIOLANTE. A. & VIOLANTE P. ( 1982): Eolian influence on terra rossa soils of Italy traced by oxy- gen isotopic ratio - In: VAN OLPHEN, H. & VENIALE, F. (eds.): 7th Int. Clay Conf. Pavia. Italy. 293-301.

JAHN. R.. ZAREI, M. & STAHR. K. (1991): Genetic implications of quartz in "Terra Rossa" soils in Portugal - Proceedings of 7th Eu- roclay Conference, Dresden, 541-546.

JOHNS. W.D., GRIM. R.E. & BRADLEY. W.F. (1954): Quantitative estimations of clay minerals by diffraction methods.- Jour. Sed. Petr.. 24.242-251.

KARATHANASIS. A.D. & HAJEK. B.F. (1984): Evaluation of allumi- num-smectite stability equilibria in naturally acidic soils.- Soil Sei. Soc. Am. J..48.413M17.

KISPATIC, M. (1912): Bauxites des Kroatischen Karstes und ihre Ein- stehung.-N. Jb. Min. Geol. Pal.. 34. 513-552.

KUBIËNA, W.L. (1953): The Soils of Europe.-Thomas Murby and Co.. London, 317 p.

KURTOVIC, J. (1973): Prilog poznavanjufaktorakoji cinecrvenice naj- boljim duhanskim tlima u Hercegovini, Simpozij o inrodukciji i proizvodnji krupnolisnih i visokoprinosnih sorata duhana u SFRJ, 173-191, Duhanski Institut, Mostar

KURTOVIC, J. (1979): Agropedoloska studija reona proizvodnje du- hana u Citluku sa pedoloskom kartom i kartom plodnosti tla, Zbornik radova, Duvanski Institut, Mostar, 182 p.

MACLEOD, D.A. (1980): The origin of the red Mediterranean soils in Epirus, Greece-J. Soil Sei., 31, 125-136.

MARIC, L. (1964): Terra Rossa u karstu Jugoslavije-Acta geológica, 4, 19-72, Zagreb.

MERINO, E. & BANERJEE, A. (2008): Terra rossa genesis, implica- tions for karst, and eolian dust: A geodynamic thread.- The Journal of Geology, 116, 62-75.

MOJICEVIC. M. & LAUSEVIC. M. (1971): Osnovna geoloska karta SFRJ 1:100.000. List Mostar K33-24 [Basic Geological Map of SFRY 1:100,000. The Mostar sheet].- Geoloski zavod Sarajevo (1958-1969). Savezni geoloski zavod. Beograd.

MOJICEVIC. M. & LAUSEVIC. M. (1973): Osnovna geoloska karta SFRJ 1:100.000. Tumac za list Mostar K33-24 Basic Geological Map of SFRY 1:100 000. Geology of the Mostar sheet.- Geoloski zavod Sarajevo. Savezni geoloski zavod Beograd. 48 p.

MOORE. D.M. & REYNOLDS. R.C. (1989): X-ray diffraction and the identification and analysis of Clay Minerals.- Oxford University Press. Oxford. 326 p.

MOREST M. & MONGELLI. G. (1988): The relation between the terra rossa and the carbonate-free residue of the underlying limestones and dolostones in Apulia. Italy.- Clay Minerals. 23. 439^146.

MUHS. D.R.. BUDHAN. J.R.. PROSPERO. J.M.. SKIPP. G. & HERWITZ.S.R. (2012):Soil genesis on the island of Bermuda in the Quaternary: The Importance of African Dust Transport and Dep- osition.- J. Geophys. Res.. 117. F03 025. doi.: 10.1029/2012 JF 002366.

MUNSELL SOIL COLOR CHARTS (1994): Macbeth Division of Koll- morgen Instruments. New Windsor. New York. USA.

NESBITT. H.W. & YOUNG. G.M. (1982): Early Proterozoic climates and plate motions inferred from major element chemistry of lu- tites-Nature. 299. 715-717.

NIHLEN. T & OLSSON. S. (1995): Influence of eolian dust on soil for- mation in the Aegean area - Zeitschrift für Geomorphologie. 39/3. 341-361.

OLSON. C.G.. RUHE. R.V. & MAUSBACH. M.J. (1980): The terra rossa limestone contact phenomena in Karst - Southern Indiana Soil Sei. Soc. Am. J.. 44. 1075-1079.

OTTNER. E. SEDOV. S.. ODBATAR. U. & WRIESSNIG. K. (2013): Grain size and mineralogical indicators of weathering in the Ober- laab loess-paleosol sequence. Upper Austria.- Quaternary Science Journal. 62. No 1. 34^13.

PLASTER. R.W. & SHERWOOD. W.C. (1971): Bedrock weathering and residual soil formation in Central Virginia.- Geol. Soc. Am. Bull.. 82. 2813-2826.

RAIC. V. PAPES. J.. BEHILOVIC. S.. CRNOLATAC. I.. MOJICEVIC. M.. RANKOVIC. M. SLISKOVIC. T. BORBEVIC. B.. GOLO. B.. AHAC. A.. LUBURIC. P. & MARIC. LJ. (1975): Osnovna geoloska karta SFRJ 1:100.000. List Metkovic K33-36 Basic Geo- logical Map of SFRY 1:100 000. The Metkovic sheet.- Geoloski zavod Sarajevo (1958-1971). Savezni geoloski zavod. Beograd.

RAIC. V. & PAPES. J. (1977): Osnovna geoloska karta SFRJ 1:100.000. Tumac za list Metkovic K33-36 (Basic Geological Map of SFRY 1:100 000. Geology of the Metkovic sheet).- Geoloski zavod Sara- jevo. Savezni geoloski zavod Beograd. 39p.

RANGE. K.J.. RANGE. A. & WEISS. A. (1969): Fire-clay type kaoli- nite or fire-clay mineral? Experimental classification of kaolinite- halloysite minerals. Proc. 3rd Int. Clay Confi. Tokyo 1. 3-13.

RAPP. A. (1984): Are terra rossa soils in Europe eolian deposits from Africa?- Geologiska Foreninges et Stockholm Forhandlingar. 105. 161-168.

RYAN. P.C.. HILLIER. S. & WALL. A4. (2008): Stepwise effects of the BCR sequential chemical extraction procedure on dissolution and metal release from common ferromagnesian clay minerals: a com- bined solution chemistry and X-ray powder diffraction study- Sei. Total Environ. 407. 603-614.

RESULOVIC. H.. ZIVANOV. N. & JOVANDIC. P. (1963): Karakteri- stike sastava adsorptivnog kompleksa na jednoj razvojnoj seriji tla na trijadicnom krecnjaku. Zemljiste i biljka. vol. XII. no. 1-3. Be- ograd.

SHELDON. N.D. (2006): Abrupt chemical weathering increase across the Pennian-Triassic boundary.-Palaeogeogr. Palaeoclimatol. Pal- aeoecol. 231.315-321.

SHELDON. N.D. & TABOR. N.J. (2009): Quantitative paleoenviron- mental and paleoclimatic reconstruction using paleosols - Earth- Sci. Rev.. 95/1-2. 1-52.

SOIL SURVEY STAFF ( 1975 ): Soil Taxonomy: A Basic System of Soil Classification for Making and Interpreting Soil Surveys.- USDA HandbookNo. 436. U.S. Government Printing Office. Washington. D.C.

SRODON. J. (1984): X-ray powder diffraction identification of illitic materials - Clays Clay Miner.. 32. 337-349.

SRODON. J. & EBERL1. D.D. (1984): Illite.- In: BAILEY. S.W. Ed.. Micas. Rev. Miner. 13. Miner. Soc. America. 495-544.

STILES. C.A. & STENSVOLD. K.A. (2008): Loess contribution to soils forming on dolostone in the Driftless Area of Wisconsin - Soil Sei. Soc. Am. J. 72. 650-659.

SINKOVEC. B. (1974): Porijeklo terra rosse Istre - Geoloski vjesnik. 27. 227-237.

SKORIC. A. FILIPOVSKI. G. & CIRIC. M. (1985): Klasifikacija tala Jugoslavije. Posebna izdanja. 13. Akademija nauka i umjetnosti BiH. Odj. prirodnih i matematickih nauka. Sarajevo.

SKORIC. A. (1979): Dvoslojni profili tla napodrucju terra rosse ulstri.- Zemljiste i biljka. 28. 111-131. Beograd.

SKORIC. A. (1987): Pedosfera Istre - Project Council of Pedological Map of Croatia. Special ed.. 2. Zagreb. 192 p.

TASSIER. A.. CAMPBELL. P.C.G. & BISSON. M. (1979): Sequential extraction procedure for the spéciation of particulate trace metals.- Analytical Chemistry 5. 844-851.

TISLJAR. J.. VLAHOVIC. T. VELIC. I. & SOKAC. B. (2002): Carbon- ate Platform Megafacies of the Jurassic and Cretaceous Deposits of the Karst Dinarides - Geol. Croat.. 55/2. 139 - 170. Zagreb.

TUCAN. F. ( 1912): Terra Rossa. derenNatur and Entstehung - Jahrbuch Min. Geol. Pak. XXXIV Beilage. 401^130.

TVICA. M. (2008): Stanje organskog karbona u vaznijim tipovima zemljista zavisno od nacina koristenja. Magistarski rad. Univerzitet u Sarajevu. 89 p.

VLAHOVIC. I. TISLJAR. J.. VELIC. I. & MATICEC. D. (2005): Evo- lution of the Adriatic Carbonate Platform: Palaeogeography. main events and depositional dynamics - Palaeogeog.. Palaeoclim.. Pal- aeoecol. 220. 3^1; 333-360.

YAALON. D.H. & GANOR. E. (1973): The influence of dust soils dur- ing the Quaternary- Soil Science. 116. 233-251.

WILSON. M.J. (2004): Weathering of the primary rock-forming miner- als: processes, products and rates - Clay Miner. 39/3. 233-266.

WRIGHT. W.R. & FOSS. J.E. (1972): Contributions of clay and organ- ic matter to the cation exchange capacity of Maryland soils - Soil Sei. Soc. Am. J. 36/1. 115-118.

Manuscript received June 20, 2014

Revised manuscript accepted September 16, 2014

Available online October 31, 2014