Geochemical, REE and Multivariate Statistical Constraints on Fe–Mn Mineralization in Durmuştepe Area (Maden, Elazığ, Türkiye)

Abstract

In this study, we examined the mineralogy, geochemistry and origin of Fe–Mn mineralization occurring as lenses within siliceous mudstones of the Middle Eocene Maden Complex (Durmuştepe, Elazığ, Türkiye) using multivariate statistical methods and Compositional Data Analysis (CoDA). Major-oxide, trace-element and REE analyses were performed on 21 samples (n = 21), comprising 11 ore and 10 host-rock samples. Ore microscopy, discrimination diagrams, PAAS-normalized REE patterns, Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA), factor analysis, Kendall correlation and CLR-transformed CoDA-PCA were applied. The ore consists of pyrolusite, braunite, manganite, magnetite and hematite. Massive and felty textures indicate rapid precipitation linked to abrupt physicochemical changes near the vents. Fe/Mn ≈ 1.62 and Co/Ni = 0.03–0.04 support an exhalative origin; Ce/Ce* = 0.10–0.15 (mean 0.11) records a pronounced negative Ce anomaly; and Eu/Eu* = 1.06–1.18 suggests vent-fluid temperatures below ~250 °C. CoDA-PCA shows that Ce is decoupled from the other lanthanides; HCA separates ore from host-rock populations; and factor analysis indicates that Ti–Al–K detrital input diluted ore accumulation. Integrating these results, we interpret Durmuştepe as a proximal volcano-sedimentary hydrothermal-exhalative Fe–Mn system that formed during magmatism in the Maden marginal basin under oxic seawater mixing and contemporaneous detrital sedimentation. The multivariate workflow also provides a reproducible approach for source characterization in similar mixed-origin deposits.

1. Introduction

Manganese deposits are divided into three main groups according to their mineralogical composition and tectonic setting: hydrogenetic (precipitation from seawater), diagenetic (precipitation from sedimentary pore water) and hydrothermal (precipitation from low- to medium-temperature hydrothermal fluids) [1,2,3]. Early models for Fe–Mn accumulations in oceanic and active-ridge settings emphasized seafloor volcanism and ridge sedimentary processes [4,5]. These deposits are also classified as oceanic or sedimentary in origin, depending on geological processes such as surface weathering, sedimentation, volcanism, plutonism and metamorphism [6,7,8,9,10,11,12]. In general, manganese deposits are systems that contain various oxide minerals and vary over a wide range in terms of grade, size and genetic origin [13,14]. On a global scale, manganese deposits are also systematically classified into sedimentary, volcano-sedimentary, hydrothermal-exhalative and similar genetic types [15]. Large industrial-scale manganese deposits are concentrated in particular stratigraphic and tectonic belts worldwide. These include the Nikopol deposit in Ukraine and the Chiatura deposit in Georgia, the Groote Eylandt deposit in northern Australia [16,17], the hydrothermal manganese deposits of Japan [18], the Tethyan deposits of Türkiye [19]—such as the limestone-hosted hydrothermal manganese mineralizations with hydrogenic contributions in Gümüşhane [20]—and the deposits of India [21]. In addition, global-scale manganese ore provinces such as the Kalahari Basin in the Republic of South Africa, the Oligocene basins of the Paratethys (Ukraine, Georgia, Kazakhstan and Bulgaria), the Northern Ural region of Russia and the Gulf of Carpentaria are well known in the literature [22,23].

Manganese precipitates on the ocean floor show distinct mineralogical differences depending on the oxidation potential of the environment. In oxygenated (oxic) pelagic settings, manganese is in the +4 oxidation state (Mn⁴⁺) and precipitates in oxide form as polymetallic nodules or crusts. By contrast, under reducing, restricted-basin conditions with low oxidation potential, manganese is in the +2 oxidation state (Mn²⁺) and, together with the Fe²⁺ form of dissolved iron, may be deposited in carbonate or silicate phases (rhodochrosite and rhodonite, respectively). Sedimentary deposits, on the other hand, form when manganese is transported as carbonate, chloride or sulfate complexes in neutral to slightly acidic waters under suitable redox/pH conditions and precipitates upon oxidation of the environment. Such deposits may have very large reserves but can contain high iron grades and clastic impurities. The main ore minerals in sedimentary-hydrothermal deposits are pyrolusite, psilomelane and rhodochrosite [24]. The most reliable approach to the genetic classification of manganese deposits is the integrated evaluation of the mineralization process and paleo-genetic environmental conditions together with mineralogical and multivariate geochemical data. The fundamental aim of this approach is to establish the genetic link between the depositional environment and the identified manganese oxide/carbonate mineralogy, and to quantitatively explain trace-element (REE and noble-metal) enrichments of hydrothermal, diagenetic or hydrogenetic origin [25,26].

Volcano-sedimentary hydrothermal-exhalative Fe–Mn deposits represent a distinct class of marine mineralizations associated with high-heat-flow volcano-tectonic systems, such as mid-ocean ridges, back-arc basins, intra-continental rifts, and island arc environments [27,28,29,30]. These deposits typically form where convective seawater circulation through the oceanic crust is driven by underlying magmatic heat sources, leaching transition metals from the basement basaltic rocks. Upon discharge onto the oxygenated seafloor, the hydrothermal fluids mix with cold seawater, triggering the rapid precipitation of Fe and Mn oxides and oxyhydroxides [31]. Morphologically, volcano-sedimentary Fe–Mn orebodies typically occur as stratabound lenses, stratiform beds, or irregular massive bodies intercalated with volcaniclastic sediments, radiolarian cherts, and pelagic mudstones [32]. At the macro scale, the ores display typical textures such as massive, layered, banded, colloform, and botryoidal structures, which are indicative of rapid deposition under fluctuating physical–chemical conditions. Microscopically, they are characterized by micro-colloform, cryptocrystalline, felty, and dendritic textures, reflecting colloid precipitation and low-temperature diagenetic recrystallization within seafloor sediments [32,33].

Geochemically, volcano-sedimentary hydrothermal Fe–Mn deposits are characterized by extreme fractionation between iron and manganese, low concentrations of cobalt, nickel, and copper (due to rapid accumulation rates preventing scavenging from seawater), and diagnostic rare earth element (REE) signatures [27,28,29]. Hydrothermal Fe–Mn precipitates typically exhibit strongly fractionated chondrite-normalized REE patterns, characterized by light REE (LREE) depletion or enrichment depending on clastic contribution, weak-to-strong negative cerium (Ce) anomalies under oxic bottom waters, and distinct positive europium (Eu) anomalies indicating high-temperature (>250 °C) fluid–rock interactions [31,32]. In contrast, hydrogenetic crusts are enriched in Co, Ni, and Cu, and exhibit strong positive Ce anomalies and negative Eu anomalies [27,28]. To distinguish between these genetic types, binary and ternary discrimination diagrams (e.g., Fe–Mn–(Co + Ni + Cu) and Fe/Ti vs. Al/(Al + Fe + Mn)) are widely applied as robust metallogenic indicators [27,33]. These geochemical signatures are often modified by suboxic diagenesis in restricted basins, leading to mineral dissolution, recrystallization, and migration of critical metals within pores and fractures [28,33].

On a global scale, manganese is primarily consumed in the iron and steel sector, followed by the chemical and battery industries [34,35,36,37,38]. Ores with high manganese dioxide content, particularly high-grade pyrolusite, serve as crucial raw materials for these metallurgical and industrial applications [34].

The Maden Complex, which lies on the Southeast Anatolian Thrust Belt in Türkiye and forms the upper part of the ophiolitic sequence, hosts numerous manganese deposits of metallogenic significance. These deposits lie within a broad tectonostratigraphic belt that extends from Kahramanmaraş in the west, through Adıyaman, Malatya and Elazığ, to Bingöl in the east, and many of them are sites that were mined in the past or still hold economic potential [3,39,40]. Consequently, characterizing the mineralogical and geochemical features of the Fe–Mn mineralizations within the Maden Complex by modern analytical methods is important not only for understanding the evolution of regional mineralization, but also for elucidating metallogenic processes in similar ophiolitic/hydrothermal systems. New approaches such as geological big data and transfer learning are also increasingly being used in manganese exploration [41].

The main aim of this study is to determine the formation mechanism, depositional environment and regional metallogenic position of the Fe–Mn mineralization within the Maden Complex (Durmuştepe, Elazığ). Within this scope, whether the mineralization is the product of a hydrothermal-exhalative system is discussed, and the roles of hydrogenetic, diagenetic and hydrothermal processes in ore formation were assessed using geochemical data. To this end, major-oxide, trace-element and rare earth element (REE) data (PAAS-normalized patterns, Ce and Eu anomalies) were evaluated together to investigate fluid temperature, redox conditions and the degree of seawater interaction. As part of the statistical evaluation of the geochemical dataset, box plots, PCA and dendrogram analyses were applied, as well as Compositional Data Analysis (CoDA)-based PCA, Kendall correlation, regression and factor analyses. The statistical and geochemical results obtained were integrated with the field-geology findings to propose a genetic model explaining the temporal–spatial evolution of the Durmuştepe deposit.

2. Geological Setting

2.1. Regional Geological Setting

Two parallel, NE–SW-trending ophiolite belts crop out in Southeast Anatolia [42,43]. The northern belt lies between the Keban–Malatya metamorphics of the Tauride platform and the Bitlis–Pütürge massifs, and comprises the Göksun–Berit (Kahramanmaraş), Meydan (Kahramanmaraş), İspendere (Malatya), Kömürhan–Guleman (Elazığ) and Killan (Diyarbakır) ophiolites [42,43,44]. These Late Cretaceous ophiolite slices accreted to the Tauride platform are reported to have supra-subduction zone (SSZ)-type oceanic lithosphere characteristics [45]. The southern ophiolite belt is located between the Bitlis–Pütürge massifs and the Arabian Platform and, towards the west, comprises the Koçali (Adıyaman) Complex—which also includes the Troodos (Cyprus) and Tekirova ophiolites—together with the Amanos, Kızıldağ (Hatay) and Baer-Bassit (Syria) ophiolites [42].

The final continent–continent collision between the Arabian and Anatolian plates occurred during the late Serravalian (end of the Middle Miocene) [46]. During this period, the Late Cretaceous Kömürhan ophiolite, the Paleozoic Pütürge metamorphics, the Late Permian–Late Cretaceous Bodrum nappe, and the Early–Middle Eocene Maden Complex, all belonging to the Tauride belt, were sliced and emplaced as nappe structures onto the northern margin of the Southeast Anatolian Autochthon [46], while similar Late Cretaceous ophiolites were emplaced onto the Arabian Platform [45,47]. These slices are considered to be remnants of the Berit Ocean [42] or the Pamphylia–Bitlis Ocean [48]. Within this tectonic framework, the Eocene Maden Complex volcanics were generated [49], and their geochemical signatures can be analyzed using trace-element discrimination approaches [50] to support the interpretation of the hydrothermal-exhalative processes that controlled the Fe–Mn mineralizations [39].

The Guleman ophiolite and the unconformably overlying flysch-type Hazar Formation represent the pre-Eocene basement and cover sequences in the region, containing various ophiolitic mélange units and volcanic intercalations [47,51,52]. Within the Maden Complex, ophiolitic slices occur as tectonic blocks [53,54,55], and their genetic origin and accretionary history have been widely discussed in the literature [56,57,58].

2.2. Local Geological Setting

The study area lies within the Eastern Taurus Orogenic Belt, where units of the Maden Complex together with the Guleman and Hazar groups crop out predominantly. From oldest to youngest, the stratigraphic succession in the study area consists of the Jurassic–Cretaceous Guleman Group, the Maastrichtian–Early Eocene Hazar Group, the Middle Eocene Maden Complex, and Pliocene–Quaternary deposits together with recent alluvium. The Simaki Formation overlies the Guleman Group along a tectonic contact in the vicinity of Hatun Köy in the northeast of the study area. The recent alluvium, the youngest unit in the region, unconformably overlies all of these sequences (Guleman, Hazar and Maden).

The Hazar Group, which transgressively overlies the Guleman Group, generally consists of a sandstone–shale–marl–limestone alternation. Conglomerates occur at the base of the unit, passing upward into sandstone, mudstone, claystone, marl, argillaceous limestone, silexite, radiolarite and pink-to-red limestones. This fossiliferous unit, referred to as the ‘lower flysch’, also contains volcanic intercalations such as basalt, diabase, melaphyre, tuff and andesite [59].

The Maden Complex covers a broad tectonic zone that extends in an approximately NE–SW direction, parallel to the East Anatolian Fault (EAF). This zone extends from the northern parts of the Palu and Arıcak districts in the east to the south of Malatya and the northern parts of Adıyaman. The Maden Complex shows more extensive outcrops in the southern parts of the fault between Palu and Lake Hazar, whereas from the southwest of Lake Hazar onward it becomes dominant in the northern blocks of the fault.

The Maden Complex in the study area shows a lithologically highly heterogeneous structure. The unit generally consists of red, green and grey silicified mudstones, maroon limestones with volcanic intercalations, a sandstone–shale–marl alternation, conglomerates, green limestone blocks, and basalts and/or spilites. The ferruginous manganese (Fe–Mn) mineralizations examined here also crop out, like the other volcanic and sedimentary units, directly as part of the lithological assemblage within the Maden Complex.

The regional tectonic positions of these units and a location map of the study area are presented in Figure 1. This map shows the tectonic relationships between the Maden Complex and the ophiolitic units and the locations of the Fe–Mn mineralizations in the region.

3. Materials and Methods

3.1. Sampling, Petrography, and Geochemical Analysis

Major-oxide, trace-element and rare earth element (REE) analyses were carried out on a total of 21 rock samples collected from the ore outcrops and host rocks in the study area. Thin sections were prepared to determine the petrographic characteristics of the geological units in the field, and polished sections to identify the mineral paragenesis and the opaque minerals. The polished sections were examined under the microscope at the laboratories of the MTA Central Anatolia Regional Directorate (Konya), and the thin sections at Fırat University. Through ore microscopy, the mineral assemblages, paragenetic relationships, and the structural and textural characteristics of the ore were described in detail.

Eleven ore samples were collected from accessible exposures of the mineralized horizon shown in Figure 1. Although approximately ten distinct outcrop/locality points are visible on the map, systematic grid sampling was not feasible because the ore occurs as discontinuous lenses within siliceous mudstones and access was limited to outcrop-scale exposures. The sampling design therefore targeted representative ore and host-rock geochemistry rather than spatial geostatistical interpolation.

Prior to the geochemical analyses, the rock samples were crushed, quartered and ground to a fineness of 125 mesh (120 µm) and homogenized. All analytical measurements were carried out at Acme Analytical Laboratories Ltd. (Vancouver, BC, Canada) in September 2013. Major-oxide analyses were performed by the 4A-4B (ICP-ES) method, trace-element and REE analyses by the 1DX (ICP-MS) method, and total carbon and sulfur analyses by the 2A (LECO) method. The accuracy and precision of the analytical data were monitored within the laboratory’s standard quality assurance/quality control (QA/QC) procedures using certified reference materials (CRM), blank samples and internal duplicates. The analytical results obtained were confirmed to lie within the laboratory’s accepted analytical error limits (relative error < ±5%).

Stable-isotope analyses (δ¹⁸O, δ¹³C) and radiogenic Sr–Nd isotope measurements were not performed in this study. The present dataset comprises whole-rock major-oxide, trace-element and REE analyses obtained from Acme Labs (September 2013). Fluid-source constraints are therefore based on trace-element and REE proxies (including Ce/Ce*, Eu/Eu*, PAAS-normalized patterns and discrimination diagrams) rather than direct isotopic fingerprinting. Isotopic work on comparable Maden Complex Fe–Mn systems remains a worthwhile direction for future research but was beyond the scope and analytical budget of the present reconnaissance-scale sampling program.

3.2. Geochemical and Statistical Methods

In evaluating the geochemical data obtained, graphical modeling and multivariate statistical analyses were used in a mutually supporting manner. First, in order to better understand the geochemical anomalies and redox conditions in the region, the rare earth element (REE) concentrations were normalized to and evaluated against the Post-Archean Australian Shale [60]. In addition, trace-element discrimination diagrams (Zn–Pb, (Co + Ni)–(As + Cu + Mo + Pb + Zn + V), Al/(Al + Fe + Mn)–Fe/Ti and Ni/Co–V/(V + Ni)) were used to determine the detrital input to the environment, to reveal the redox (oxic/anoxic) conditions during deposition, and to identify whether the origin of the mineralization involved hydrothermal, hydrogenetic or diagenetic processes.

The statistical investigations of the geochemical dataset of the Durmuştepe Fe–Mn mineralization were carried out in a two-stage process.

Descriptive statistics and box plots were generated using XLSTAT (v19.2.2) and CorelDraw software (Graphics Suite v26.2.0.170). Hierarchical Cluster Analysis (HCA) and Principal Component Analysis (PCA) were performed using SPSS v26 software to identify sample groups and distinguish ore from host rocks. Prior to PCA, variables were normalized using Z-score standardization because the dataset contains variables measured in different units and scales (major oxides in wt.% and trace elements/REEs in ppm). Without standardization, elements with high concentrations (e.g., SiO₂ and Fe₂O₃) would dominate the variance. Z-score standardization scales each variable to a mean of 0 and a variance of 1 using the formula:

Zi = (Xi − μ)/σ,

where Xi is the raw concentration, μ is the mean, and σ is the standard deviation.

HCA was performed using Euclidean distance and Ward’s method. For PCA, a Pearson correlation matrix was computed from the standardized variables (equivalent to the covariance matrix of Z-scores). Eigenvalues and eigenvectors were extracted to obtain the principal components and variable loadings; component scores and loading vectors were displayed together in a biplot. Principal components were retained according to Kaiser’s rule (eigenvalues > 1), cumulative explained variance, and inspection of the scree plot. This standardization and PCA approach follows methods described for geochemical classification and mineral characterization [61,62].

In the second stage, evaluations were carried out on the ore samples (n = 11). The Shapiro–Wilk test, applied to determine the distribution characteristics of the elements in these samples, revealed that the elements were not normally distributed (p < 0.05). Therefore, in the correlation analyses, the Kendall correlation—which does not require an assumption of normality and is insensitive to outliers—was preferred. Kendall’s correlation was selected over Spearman’s rank correlation because it is mathematically more robust for small sample sizes (n = 11), exhibits smaller gross-error sensitivity, and provides more reliable p-values and confidence intervals under non-normal distributions with extreme values [63]. Because the number of ore samples was limited (n = 11), the statistical results were evaluated at a descriptive and interpretive level.

Given the limited number of ore samples (n = 11), all multivariate analyses (PCA, HCA, factor analysis, Kendall correlation, and CoDA-PCA) were treated as exploratory tools rather than confirmatory models. With more than 30 geochemical variables and only 11 observations, the ratio of observations to variables is low; therefore, the extracted components and factors should not be interpreted as statistically robust, independent end-members. Instead, they were used to identify broad geochemical trends and to support—qualitatively—the interpretations derived from ore microscopy, discrimination diagrams, and REE patterns. Interpretive statements based on multivariate outputs were deliberately framed in non-causal terms (e.g., “suggests”, “is consistent with”) and were not used as standalone proof of genetic processes.

Compositional Data Analysis (CoDA) was performed using CoDaPack version 2.03.11 software on the compositional geochemical data [64,65]. A centered log-ratio (CLR) transformation was applied to the ore data; prior to the transformation, zero and very low values were replaced with positive values. Correlation and regression analyses were carried out using OriginPro 2025b v10.2.5.212 software, and factor analyses were also carried out for the major-oxide, trace-element and REE groups. Varimax rotation was applied to improve the interpretability of the factor loadings. Varimax orthogonal rotation was preferred over oblique rotation to identify clear, independent geochemical end-members (e.g., pure hydrothermal-exhalative inputs versus detrital dilution processes). Oblique rotations (such as Promax) were also tested and yielded highly consistent factor groupings, supporting the exploratory use of the orthogonal model in isolating these distinct processes [66].

For the revised factor analysis, all 44 geochemical variables (major oxides, trace elements and REEs) were used for the eleven ore samples (Fe-4, Fe-5a/b, Fe-7a/b, Fe-9a, Fe-11, Fe-12a/b, Fe-13, Fe-14). Variables were Z-score standardized prior to extraction. Factors were extracted using principal-component-based factor analysis with Varimax rotation, consistent with the original SPSS/OriginPro workflow. Although the Kaiser criterion initially suggested six factors (eigenvalues > 1), inspection of the scree plot and parsimony for small-n exploratory interpretation led to retention of four factors (cumulative explained variance = 99.8%; see Section 4.5.3).

The conceptual genetic model figure was prepared by the authors. During the preparation of this work, the authors used Google NotebookLM solely to assist with the initial graphical layout and schematic visualization of this figure. The authors’ own manuscript data and selected regional geological publications were used as input. The final vector drafting, process pathways, flow arrows, annotations and geological interpretation were completed manually by the authors in Corel Draw. Generative artificial intelligence tools were also used to assist with English-language editing and reference formatting. All AI-assisted outputs were reviewed, corrected, and approved by the authors. No generative artificial intelligence tools were used for geochemical data analysis or statistical processing.

4. Results

4.1. Ore Mineralogy and Petrography

In order to determine the textural relationships and mineralogical paragenesis observed in the Durmuştepe Fe–Mn mineralization, polished-section studies were carried out on samples collected from the ore outcrops and the surrounding host rocks. The microscopic examinations showed that pyrolusite, braunite, manganite, magnetite and hematite constitute the main ore phases, while pentlandite [(Ni,Fe)₉S₈] and copper minerals (cubanite, malachite, azurite) were detected locally.

At the outcrop scale, the stratabound Fe–Mn mineralization occurs in sharp contact with the host mudstone (Figure 2A), showing clear thickness variations along a fault-controlled structural contact (Figure 2B). Hand specimens of the high-grade, massive manganese ore display a characteristic steel-grey metallic luster and a dense, blocky structure (Figure 2C), contrasting with the highly weathered, reddish-brown limonitic clay and silica-cemented mudstone host rocks that border the mineralization zone (Figure 2D). Observations at the macroscopic and microscopic scales indicate that the mineralization generally displays a massive character, while in some zones fractured–cracked, porous and felty textures have developed. The oxide phases that make up the ore consist mostly of subhedral and anhedral crystals and display a pervasive intergrown texture. Particularly in the fine-grained (15–20 μm) zones with felty texture, it becomes difficult to distinguish the manganese minerals individually under the microscope. This fine-grained, porous matrix is interpreted to consist of an intimate mixture of pyrolusite, psilomelane, rhodochrosite and silicate phases. In the massive parts, by contrast, manganite, braunite and pyrolusite were found to be clustered as coarser-grained aggregates.

The most commonly observed phase among the manganese minerals in the paragenesis is pyrolusite (MnO₂). This mineral is readily distinguished from the other phases by its cream-white color, high reflectance and distinct anisotropy (Figure 2E,G). The pyrolusite crystals, whose sizes vary considerably, range from millimeter-scale (~2 mm) grains to micron-scale fine components. The coarse-crystalline pyrolusite that forms the felty aggregates is thought to represent the primary mineralization stage, whereas the fine-grained pyrolusite filling the cracks and fractures (locally accompanied by psilomelane) is thought to have developed through secondary processes. Braunite (3Mn₂O₃·MnSiO₃), characterized by its low anisotropy and brownish-grey color tones, mostly occurs in paragenetic association with manganite (MnO(OH)) and pyrolusite and is among the main components of the massive matrix. Manganese-magnetite (Mn-Mt) phases are also observed locally, emplaced within microcrack and vein systems cutting the massive matrix (Figure 2F). These Mn-Mt phases do not represent the main ore phase, which is dominated by braunite and pyrolusite, but rather reflect a transitional or late-stage hydrothermal deposition. Furthermore, fine-grained secondary manganese oxides occur as disseminations filling voids within the silica-cemented mudstone matrix (Figure 2H), demonstrating secondary mineral enrichment during post-depositional fluid circulation.

Among the iron-oxide phases accompanying the mineralization, magnetite and hematite were observed intensely, especially in the gabbro-type host rocks belonging to the Guleman Group. At the margins and in the microcracks of the magnetites in the gabbros, conversion to hematite (martitization texture) developed under the influence of oxidation processes is clearly observed (Figure 2I). In these mafic units, yellow-brown, metallic-lustered pentlandite crystals representing the sulfide phase display a close paragenetic relationship with magnetite (Figure 2J). When the microscopic data are evaluated as a whole, it was determined that the Fe–Mn mineralization in the study area underwent a multi-stage evolution, with secondary fracture fillings and disseminated phases developing in the silicified mudstones after the main massive mineralization stage.

4.2. Whole-Rock Geochemistry and Descriptive Statistics

The statistical evaluations in this study were carried out on the trace-element (Au, Co, Hf, Nb, Ta, V, Zr, Y, Mo, Cu, Pb, Zn, Ni, As, Hg, Ba), rare earth element (La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu) and major-oxide (SiO₂, Al₂O₃, Fe₂O₃, MgO, CaO, Na₂O, K₂O, TiO₂, MnO, Cr₂O₃) analytical results of 11 ore and 10 host-rock/alteration samples (total n = 21) collected from the Durmuştepe Fe–Mn mineralization area (

Table 1. Major-oxide contents (wt.%) of rock samples from the Durmuştepe study area. (S. No.: Sample Number, Mean: Arithmetic Mean, U. L.: Upper Limit, L. L.: Lower Limit, Std. Dev.: Standard Deviation, LOI: Loss on Ignition, TOT/C: Total Carbon, TOT/S: Total Sulfur).

S. No.	Sample Type	SiO2	Al2O3	Fe2O3	MgO	CaO	Na2O	K2O	TiO2	P2O5	MnO	Cr2O3	LOI	TOT/C	TOT/S	Fe	Mn	Fe/Mn
Fe-5a	Ore sample	18.03	2.11	41.28	0.29	4.93	0.05	0.03	0.12	1.04	20.65	0.006	11	0.04	0.02	28.87	16	1.80
Fe-5b	Ore sample	18.36	2.18	41.47	0.36	5.07	0.05	0.03	0.12	1.01	19.85	0.006	11	0.04	0.02	29	15.37	1.89
Fe-7a	Ore sample	25.59	6.55	27.28	0.57	16.2	0.09	0.04	0.2	0.65	16.47	0.004	5.8	0.06	0.02	19.08	12.76	1.50
Fe-7b	Ore sample	24.71	3.15	29.48	0.3	17	0.02	0.05	0.1	1.56	17.47	0.004	5.7	0.27	0.02	20.62	13.53	1.52
Fe-12a	Ore sample	24.41	5.29	25.85	0.41	14.9	0.09	0.33	0.12	0.75	18.02	0.015	9.2	0.19	0.02	18.08	13.96	1.30
Fe-12b	Ore sample	27.62	4.8	25.25	0.19	14.7	0.07	0.12	0.09	0.56	16.97	0.008	9.1	0.1	0.02	17.73	13.14	1.35
Fe-11	Ore sample	22.03	6.7	39.33	1.75	6.44	0.03	0.1	0.25	0.62	15.57	0.006	6.6	0.05	0.02	27.51	12.06	2.28
Fe-13	Ore sample	23.47	3.07	31.44	0.27	15	0.02	0.07	0.13	0.94	19.5	0.007	5.5	0.21	0.02	22	15.1	1.46
Fe-14	Ore sample	24.83	3.68	30.97	0.36	14.3	0.06	0.06	0.14	1.1	18.79	0.006	5.2	0.14	0.02	21.66	14.55	1.49
Fe-4	Ore sample	26.22	4.67	28.61	0.1	13.6	0.02	0.12	0.09	0.51	16.22	0.003	9.3	0.03	0.02	20.01	12.56	1.59
Fe-9a	Gabbros/altered	49.56	14.31	15.04	4.97	4.37	4.74	0.05	2.45	0.64	0.32	0.002	3.3	0.12	0.12
G1	Gabbros	57.28	14.94	9.93	2.91	2.84	5.33	0.07	1.2	0.43	0.16	0.003	4.6	0.34	0.02
M1G	Serpentinite	36.69	1.42	11.63	33.1	3.05	0.02	0.01	0.03	0.02	0.26	0.485	13	0.45	0.02
D3	Siliceous mudstone /Ore side rock	74.35	2.94	16.87	1.31	0.94	0.03	0.18	0.1	0.03	0.44	0.007	2.7	0.1	0.02
Ma-1	Malachite/Azurite-rich specimen	47.79	16.34	8.41	6.68	3.51	3.98	0.16	0.85	0.29	0.47	0.009	6.5	0.21	0.41
Ma-2	Malachite/Azurite-rich specimen	49.63	17.86	9.07	7.31	2	4.94	0.31	0.77	0.17	0.57	0.008	5.8	0.09	0.02
H5	Serpentinite	38.62	5.25	8.69	32.6	1.22	0.02	0.01	0.02	0.01	0.15	0.377	13	0.06	0.04
H9	Mudstone	50.32	14.28	8.38	4.25	7.14	0.8	3.04	0.73	0.15	0.62	0.039	10	1.47	0.02
H11	Sandstone	52.71	17.87	8.75	5.16	2.51	6.22	0.55	0.93	0.18	0.5	0.014	4.4	0.05	0.02
H12	Mudstone	57.3	14.61	10.15	1.78	5.38	5.49	0.14	0.63	0.21	0.11	0.015	3.7	0.63	0.02
H13	Sandstone	53.52	18.08	8.23	5.35	2.44	6.51	0.68	0.91	0.15	0.22	0.013	3.7	0.04	0.02
Mean		38.24	8.58	20.77	5.23	7.5	1.84	0.3	0.48	0.53	8.73	0.05	7.1	0.22	0.05	21.45	13.9	1.62
U. L.		74.35	18.08	41.47	33.05	17.01	6.51	3.04	2.45	1.56	20.65	0.49	13	1.47	0.41
L. L.		18.03	1.42	8.23	0.1	0.94	0.02	0.01	0.02	0.01	0.11	0	2.7	0.03	0.02
Std. Dev.		15.78	6.07	11.73	9.23	5.61	2.51	0.64	0.57	0.42	8.86	0.13	3.1	0.32	0.08

,

Table 2. Trace-element contents (ppm) of rock samples from the Durmuştepe study area. (S. No.: Sample Number, Mean: Arithmetic Mean, U. L.: Upper Limit, L. L.: Lower Limit, Std. Dev.: Standard Deviation).

S. No.	Sample Type	Au	Co	Hf	Nb	V	W	Zr	Y	Mo	Cu	Pb	Zn	Ni	Ba	As	Hg
Fe-5a	Ore sample	0.004	73.3	1.3	4.8	734	18.4	84.4	135.5	1.8	1796.6	74.4	370	1937.9	16	139.7	0.01
Fe-5b	Ore sample	0.004	75.9	1	4.9	768	18.1	91.9	139.2	2	1651.2	65.3	291	1920	18	140.3	0.01
Fe-7a	Ore sample	0.002	70.9	1.2	4.3	1103	6.6	71.6	103.3	6.7	1089.4	109.2	311	1889.3	204	99.9	0.02
Fe-7b	Ore sample	0.002	54.5	0.7	2.9	1283	8.4	56.9	93.9	4.6	723.5	138.2	372	1620.4	23	233.3	0.03
Fe-12a	Ore sample	0.016	61.9	0.9	3.5	1536	16.6	63.6	122.7	6.2	703.3	152.8	309	1848.2	123	94.9	0.09
Fe-12b	Ore sample	0.027	53.7	0.7	2.6	1642	13.6	52.4	73.5	4.4	780.4	184.4	372	1641	17	114.5	0.06
Fe-11	Ore sample	0.002	72	1.4	7	1166	4.5	113	122.6	1	1263.2	143	423	1700.1	81	33.6	0.01
Fe-13	Ore sample	0.001	60.8	0.9	3.4	1316	6.4	69.7	114.8	4.6	936.7	157.4	381	1609.5	29	240.7	0.07
Fe-14	Ore sample	0.002	61.8	0.9	3.4	1009	7.3	67.2	108.1	5.5	1496.7	189.5	305	1694.7	100	185.1	0.05
Fe-4	Ore sample	0.006	60.6	0.8	3.2	1667	15.2	63.5	98	2.7	981	222.4	313	1570.2	12	103.3	0.03
Fe-9a	Gabbros/altered	0.001	27.3	3.3	26	277	0.5	160	49.7	0.9	368.7	0.8	105	25.2	178	1.7	0.01
G1	Gabbros	0.004	14.7	6	54	63	0.9	277	61.4	1	888	69.2	557	33.8	48	3.1	0.01
M1G	Serpentinite	0.001	140	0.1	0.1	50	0.5	1	1.3	0.1	28.6	0.5	21	2638.7	3	4.1	0.01
D3	Siliceous mudstone /Ore side rock	0.001	11.1	0.2	0.7	197	1.7	10.3	12.6	0.5	131.9	6	36	95.5	127	2.3	0.02
Ma-1	Malachite/Azurite-rich specimen	0.036	22.3	2.5	3.8	233	0.5	106	28.5	0.7	10,000	5.2	71	73.8	89	0.9	0.1
Ma-2	Malachite/Azurite-rich specimen	0.002	26.2	2.5	3.5	241	0.5	100	30.5	0.4	10,000	2.6	79	96.2	46	2.7	0.07
H5	Serpentinite	0.001	124	0.1	0.1	29	0.5	0.8	0.6	0.1	26.6	0.3	17	2828.5	3	1.3	0.01
H9	Mudstone	0.002	44.2	3.6	15	144	1.1	126	29.7	0.4	46.6	15.9	99	476.9	286	6.2	0.07
H11	Sandstone	0.002	27.3	2.2	3.4	256	0.5	95	23.9	0.6	38.4	5.3	80	111.5	87	3.8	0.02
H12	Mudstone	0.006	31.1	1.8	4	208	0.5	70	33	2.2	1657.2	88.9	904	189.8	44	4.3	0.06
H13	Sandstone	0.001	25	2.2	3.3	237	0.5	94.2	23.3	0.6	20.6	3	73	77.4	91	2.3	0.01
Mean		0.0058	54.2	1.6	7.3	674.2	5.85	84.5	66.96	2.24	1649	77.82	261	1146.6	77	67.5	0.04
U. L.		0.036	140.2	6	54	1667	18.4	276.9	139.2	6.7	10,000	222.4	904	2828.5	286	240.7	0.1
L. L.		0.001	11.1	0.1	0.1	29	0.5	0.8	0.6	0.1	20.6	0.3	17	25.2	3	0.9	0.01
Std. Dev.		0.009	32.4	1.4	11.8	569.4	6.5	57.4	45.9	2.1	2769.2	73.9	212	931.6	72.3	79.9	0.03

,

Table 3. Rare earth element (REE) contents (ppm) of rock samples from the Durmuştepe study area (S. No.: Sample Number, Mean: Arithmetic Mean, U. L.: Upper Limit, L. L.: Lower Limit, Std. Dev.: Standard Deviation).

S. No.	Sample Type	La	Ce	Pr	Nd	Sm	Eu	Gd	Tb	Dy	Ho	Er	Tm	Yb	Lu
Fe-5a	Ore sample	171.4	34.4	36.72	155.3	27.63	6.37	27.72	4.3	24.6	4.85	13.65	1.94	11.45	1.68
Fe-5b	Ore sample	190.1	38.2	38.7	159.2	29.33	6.71	30.37	4.33	27.19	5.19	14.04	2.03	11.76	1.66
Fe-7a	Ore sample	140.4	32.9	29.7	119	22.18	5.29	22.58	3.09	19.57	3.73	9.58	1.35	7.69	1.09
Fe-7b	Ore sample	137.4	29.7	27	105.9	19.56	5.19	22.06	3.05	19.23	3.61	9.45	1.29	7.92	1.16
Fe-12a	Ore sample	141.5	31	29.16	121.8	23.48	5.9	25.74	3.53	21.9	4.13	11.03	1.58	9.28	1.31
Fe-12b	Ore sample	120	27.9	25.53	102.2	18.5	4.46	18.35	2.55	16.15	2.99	8.09	1.12	6.69	0.95
Fe-11	Ore sample	167.4	52.5	38.8	159.9	30.52	7.13	28.91	4.38	25.73	5.1	12.97	1.96	11.45	1.58
Fe-13	Ore sample	153	32.5	31.73	126.2	24.75	6.4	26.62	3.59	23.3	4.11	10.93	1.6	9.25	1.32
Fe-14	Ore sample	155.6	33.7	30.89	124.4	22.58	5.71	24.16	3.31	21.86	4.02	10.93	1.48	8.61	1.24
Fe-4	Ore sample	146.5	29.9	29.31	115	22.18	5.68	23.86	3.29	20.77	3.75	10.08	1.38	8.16	1.22
Fe-9a	Gabbros/altered	33.5	67.8	8.66	35	7.96	2.47	9.06	1.31	8.96	1.74	4.93	0.77	4.54	0.71
G1	Gabbros	71	129.1	15.09	56.4	10	2.98	10.68	1.54	11.18	2.17	6.41	1	6.38	0.91
M1G	Serpentinite	1	0.3	0.16	0.5	0.09	0.05	0.22	0.03	0.23	0.03	0.12	0.02	0.12	0.03
D3	Siliceous mudstone /Ore side rock	4.9	4.9	1.27	4.9	1.36	0.49	1.8	0.32	2.56	0.49	1.21	0.21	1.41	0.18
Ma-1	Malachite/Azurite-rich specimen	14.7	28.1	3.89	16.7	4.05	1.35	4.56	0.69	4.72	0.94	2.72	0.42	2.93	0.45
Ma-2	Malachite/Azurite-rich specimen	15.9	29.9	4.59	19.4	4.57	1.51	5.37	0.78	5.56	1.07	2.95	0.45	2.81	0.47
H5	Serpentinite	0.3	0.2	0.03	0.3	0.05	0.04	0.1	0.01	0.09	0.02	0.08	0.01	0.09	0.01
H9	Mudstone	33.4	67.1	8.57	33.1	6.29	1.54	5.97	0.81	4.99	0.93	2.64	0.38	2.36	0.36
H11	Sandstone	14.3	23.7	3.77	16.8	3.68	1.29	4.31	0.62	4.47	0.84	2.61	0.39	2.39	0.39
H12	Mudstone	26.9	35.9	6.75	27	5.79	1.61	6.34	0.91	5.85	1.11	3.01	0.48	2.92	0.47
H13	Sandstone	11	22.6	3.18	14.3	3.51	1.19	3.69	0.57	3.95	0.77	2.31	0.38	2.33	0.34
Mean		83.34	35.82	17.79	72.06	13.72	3.49	14.4	2.05	12.99	2.46	6.66	0.96	5.74	0.84
U. L.		190.1	129.1	38.8	159.9	30.52	7.13	30.37	4.38	27.19	5.19	14.04	2.03	11.76	1.68
L. L.		0.3	0.2	0.03	0.3	0.05	0.04	0.1	0.01	0.09	0.02	0.08	0.01	0.09	0.01
Std. Dev.		68.52	26.79	14.02	57.18	10.44	2.42	10.66	1.52	9.17	1.73	4.59	0.65	3.73	0.52

and

Table 4. REE parameters and ratios for Durmuştepe Fe–Mn mineralization samples (S. No.: Sample Number, Mean: Arithmetic Mean).

S. No.	∑REE	LREE/HREE	Eu/Sm	Sm/Nd	LREE/(HREE + Y)	La/Yb	Ce/La	Y/Ho	δ Ce	δ Eu	Ce (Anom)
Fe-5a	522.01	4.79	0.23	0.18	1.91	14.97	0.20	27.94	0.10	1.08	−1.00
Fe-5b	558.81	4.79	0.23	0.18	1.96	16.16	0.20	26.82	0.10	1.06	−0.99
Fe-7a	418.15	5.09	0.24	0.19	2.03	18.26	0.23	27.69	0.12	1.11	−0.93
Fe-7b	392.49	4.79	0.27	0.18	2.01	17.35	0.22	26.01	0.11	1.18	−0.95
Fe-12a	431.34	4.49	0.25	0.19	1.75	15.25	0.22	29.71	0.11	1.13	−0.95
Fe-12b	355.48	5.25	0.24	0.18	2.29	17.94	0.23	24.58	0.12	1.14	−0.93
Fe-11	548.33	4.95	0.23	0.19	2.13	14.62	0.31	24.04	0.15	1.13	−0.82
Fe-13	455.30	4.64	0.26	0.20	1.92	16.54	0.21	27.93	0.11	1.17	−0.97
Fe-14	448.49	4.93	0.25	0.18	2.03	18.07	0.22	26.89	0.11	1.15	−0.95
Fe-4	421.08	4.81	0.26	0.19	2.04	17.95	0.20	26.13	0.11	1.16	−0.98
Mean	455.15	4.85	0.25	0.19	2.01	16.71	0.23	26.78	0.11	1.13	−0.95

).

LREE/HREE = (La + Ce + Pr + Nd + Sm + Eu)/(Gd + Tb + Dy + Ho + Er + Tm + Yb + Lu); ∑Ce/∑Y (LREE/YREY) = (La + Ce + Pr + Nd + Sm + Eu)/(Gd + Tb + Dy + Ho + Er + Tm + Yb + Lu + Y); Eu/Eu* = shale-normalized (PAAS) [Eu_SN/(Sm_SN * Gd_SN)^0.5]; Ce/Ce* = shale-normalized (PAAS) [Ce_SN/(La_SN × Pr_SN)^0.5] (Note: the SN subscript in the equations denotes values normalized to the Post-Archean Australian Shale (PAAS) based on [62]. The LREE/HREE and anomaly formulations were calculated based on the standards of [67,68,69,70]). In Table 4, the columns labeled δ Ce and δ Eu report the PAAS-normalized Ce/Ce* and Eu/Eu* values, respectively. The column Ce(anom) denotes the cerium anomaly calculated as log10[3 × Ce_SN/(2 × La_SN + Nd_SN)] (values < 0 indicate a negative Ce anomaly), following the REE anomaly conventions summarized in [67,70].

4.3. Trace-Element Geochemistry and Discrimination Diagrams

Trace-element data and discrimination diagrams play an important role in the genetic interpretation of manganese deposits. The high adsorption capacity of manganese oxides allows many elements in the depositional environment to be retained by these phases. In order to interpret these elemental enrichments—which are particularly pronounced in marine manganese occurrences—correctly, the geochemical data were evaluated together with field observations and mineralogical analyses [71,72].

On the multi-metal (As + Cu + Mo + Pb + Zn + V) versus (Co + Ni) diagram (Figure 3A), which separates hydrothermal and hydrogenetic fields [73], most of the data plot in the hydrothermal field. On the Pb–Zn diagram (Figure 3C), which separates submarine hydrothermal/marine-SEDEX fields from sulfide oxidation fields (modified after [73]), the analyzed samples cluster below the oxidation boundary. Because deciding on the origin of Fe–Mn occurrences from a single diagram can yield misleading results, multi-parametric approaches are emphasized in the literature [1,26,74].

Indeed, the Fe–Mn–(Ni + Co + Cu) × 10 ternary discrimination diagram and redox-sensitive trace-element ratios are frequently used to detect hydrothermal fluid input and reconstruct paleoredox conditions in oceanic-crust-related Tethyan manganese deposits [75,76,77] and sedimentary basins [78,79,80]. Within this framework, the samples of the Durmuştepe Fe–Mn mineralization were evaluated on four different diagrams (Figure 3): (A) (As + Cu + Mo + Pb + Zn + V)–(Co + Ni), (B) Fe–Mn–(Ni + Co + Cu) × 10, (C) Pb–Zn, and (D) V/(V + Ni)–Ni/Co. In the ternary diagram (Figure 3B; after [81,82]), the ore samples cluster clearly in the hydrothermal field, reflecting a dominant exhalative signature. On the redox diagram (Figure 3D; after Rimmer [83] and Amiewalan and Lucas [84]), the ore samples cluster at high Ni/Co ratios (23–31) in the suboxic–anoxic field of the horizontal axis, whereas their V/(V + Ni) ratios (0.28–0.52) span the oxic–anoxic transition on the vertical axis. By contrast, the siliceous mudstone, mudstone, gabbro and sandstone samples plot more dispersedly, with several host samples showing higher V/(V + Ni) but lower Ni/Co than the ore cluster. This contrast reflects the difference between the bulk geochemical signature of hydrothermally precipitated Fe–Mn oxides—strongly enriched in Ni and other elements scavenged during exhalative accumulation—and the redox-sensitive ratios of the fine-grained siliceous and clastic host succession. The elevated Ni/Co of the ore samples is therefore interpreted primarily as a hydrothermal-exhalative overprint, not as evidence of uniformly anoxic bottom waters during ore formation. The negative Ce anomaly and oxide-dominated mineralogy nevertheless indicate mixing with oxygenated seawater at the seafloor exhalative interface (Section 4.4 and Section 5), consistent with the proposed proximal volcano-sedimentary hydrothermal model. The HFSE–REE discrimination schemes proposed for oceanic Fe–Mn deposits [26] and similar diagram applications [85] form the basis of this framework.

Figure 3. Geochemical discrimination diagrams of the Durmuştepe Fe–Mn mineralization and associated rock samples. (A) Hydrothermal/hydrogenetic origin discrimination using the multi-metal (As + Cu + Mo + Pb + Zn + V)–(Co + Ni) relationship [75]. (B) Hydrothermal and detrital field boundaries on the Fe–Mn–(Ni + Co + Cu) × 10 diagram [76,77]. (C) Distinction between sulfide oxidation and submarine marine–hydrothermal–SEDEX fields using the Pb–Zn relationship (modified after [75]). (D) Redox-sensitive V/(V + Ni) versus Ni/Co relationships showing oxic–dysoxic–suboxic/anoxic (Ni/Co) and oxic–anoxic–euxinic [V/(V + Ni)] fields (modified after Rimmer [83]; Amiewalan and Lucas [84]). Ore and host-rock samples are distinguished by symbol and color. The plotted data points represent the authors’ own original analytical data (this study); the classification fields and boundaries are adapted/redrawn from the cited publications.

Figure 3. Geochemical discrimination diagrams of the Durmuştepe Fe–Mn mineralization and associated rock samples. (A) Hydrothermal/hydrogenetic origin discrimination using the multi-metal (As + Cu + Mo + Pb + Zn + V)–(Co + Ni) relationship [75]. (B) Hydrothermal and detrital field boundaries on the Fe–Mn–(Ni + Co + Cu) × 10 diagram [76,77]. (C) Distinction between sulfide oxidation and submarine marine–hydrothermal–SEDEX fields using the Pb–Zn relationship (modified after [75]). (D) Redox-sensitive V/(V + Ni) versus Ni/Co relationships showing oxic–dysoxic–suboxic/anoxic (Ni/Co) and oxic–anoxic–euxinic [V/(V + Ni)] fields (modified after Rimmer [83]; Amiewalan and Lucas [84]). Ore and host-rock samples are distinguished by symbol and color. The plotted data points represent the authors’ own original analytical data (this study); the classification fields and boundaries are adapted/redrawn from the cited publications.

4.4. Rare Earth Element (REE) Geochemistry

Rare earth elements (REEs) are among the indicators frequently used to determine the genetic processes and depositional environments of Fe–Mn mineralizations [1,74,86,87]. The total REE (∑REE) contents of the Durmuştepe ore samples were found to range between 355.48 ppm and 558.81 ppm, with a mean value of 455.15 ppm (Table 4; details of all rock analyses are presented in Table 3). The REE concentrations obtained were normalized to PAAS [60] values to construct the distribution patterns (Figure 4).

PAAS-normalized REE distributions are used to distinguish hydrogenetic, diagenetic and hydrothermal origins from one another [26,88]. While a distinct positive Ce anomaly (Ce/Ce* > 1) is observed in typical hydrogenetic crust formations, strong negative Ce anomalies (Ce/Ce* < 0.5) and generally a positive Eu anomaly (Eu/Eu* > 1) are seen in hydrothermal deposits that develop through the mixing of high-temperature reducing fluids with seawater [1,74]. The redox-sensitive behavior and isotopic fractionation of cerium in marine environments are also being examined in more detail using the high-precision MC-ICP-MS analytical techniques developed in recent years [89]. Systems dominated by diagenetic processes, on the other hand, generally display more horizontal profiles and limited Ce fluctuations [74].

Moreover, the abrupt mixing and oxidation of hydrothermal fluids spreading onto the seafloor with seawater (seawater shock) leads to the fractionation of cerium, producing pronounced negative Ce anomalies (Ce/Ce* < 0.5) of the type commonly observed in deep-sea cherts. This has previously been reported as an indicator of hydrothermal solution–seawater interaction in the Vezirler oceanic manganese deposit [75], the Permian deposits of the Zunyi region [78], the Kumluca Mn occurrences [77] and the Dongping sedimentary Mn deposit [80].

When the PAAS-normalized [60] REE distribution patterns of the ore samples are examined (Figure 4), profiles that run parallel to one another and display a distinct negative Ce anomaly (Ce/Ce* ≈ 0.10–0.15; Table 4) stand out. On the multi-field distribution diagram (Figure 4; modified after [88]), the analyzed Fe–Mn samples cluster in a tight band situated between the hydrogenetic/diagenetic transition field and the Fe-rich field, confirming a mixed genetic origin for the mineralization. The negative Ce anomaly is consistent with rapid precipitation and deposition from seawater-influenced hydrothermal fluids. The Eu/Eu* values range between 1.06 and 1.18 and do not show a distinct positive Eu anomaly, which suggests that the fluid temperature remained below 250–300 °C [86,87].

In the literature, rare earth elements are generally divided into two main groups: the light REEs (LREEs) between La and Sm, and the heavy REEs (HREEs) between Gd and Lu. The element Eu is additionally treated as a transition element between these two groups [60,90]. The LREE/HREE values and the related parametric ratios in this study are compiled in detail in Table 4.

4.5. Multivariate Statistical Analysis of Geochemical Data

The statistical evaluations were carried out in accordance with the two-stage methodological framework detailed in Section 3.2. In the first stage, box plots, Principal Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA) were applied to all samples (n = 21), and it was determined that the ore and host-rock samples were geochemically well distinguished from one another. This finding demonstrated the methodological validity of restricting the subsequent statistical analyses to the ore samples only (n = 11), in order to eliminate the lithology-derived background noise (background effect) and to focus directly on the mineralization mechanism.

4.5.1. Descriptive Distribution Analysis (Box Plots)

The data distributions of the major oxides, trace elements and rare earth elements (REEs), together with possible outliers, were summarized using box plots on a base-10 logarithmic scale (Figure 5). In exploratory geochemical data analysis, this non-parametric graphical approach is widely used to separate background values from anomalies [63,91].

Among the major oxides, the highest variability and the widest distribution range are observed in SiO₂. Fe₂O₃ also shows a wide distribution, revealing the marked iron enrichment in the system. Al₂O₃, CaO and MnO display a moderate distribution, whereas the concentrations of MgO, Na₂O, K₂O and TiO₂ lie within a narrow range. These distribution characteristics point to a heterogeneous environment controlled simultaneously by detrital-material input and Fe–Mn oxy-hydroxide precipitation.

Among the trace elements, an extreme (outlier) value of copper (Cu) reaching about 10,000 ppm was detected in samples Ma-1 and Ma-2 (Table 2). These malachite- and azurite-bearing specimens were collected within the mineralization zone and are characterized by secondary Cu-carbonate mineralization (Figure 2) rather than the primary pyrolusite–braunite Fe–Mn ore assemblage; their low MnO contents (0.47–0.57 wt.%) support this interpretation. We therefore regard Ma-1 and Ma-2 as supergene/alteration products linked to the same broader hydrothermal system that formed the Fe–Mn lenses, but not as representative primary ore samples. Accordingly, they were retained in the full geochemical dataset (n = 21) for PCA, HCA and discrimination diagrams to illustrate contrast with the Fe–Mn ore cluster, but were excluded from the ore-only factor analysis (n = 11) and ore-based CoDA-PCA so that the multivariate models would not be dominated by the extreme Cu outlier.

Ni, Pb and V show wider distribution ranges and higher maximum values than the other trace elements. When the extreme Cu values of Ma-1 and Ma-2 are excluded, the narrower distribution of the other trace elements reflects the differences in the mechanisms by which the elements were incorporated into the system.

Among the rare earth elements (REEs), La and Nd from the light REE (LREE) group show high concentrations and wide distribution ranges. Although Ce displays a narrow range, it has distinct extreme values at the upper limit. The heavy REEs (HREEs) such as Er, Tm, Yb and Lu, on the other hand, are represented by low concentrations and very narrow distributions.

The box plots show the heterogeneous nature of the geochemical dataset in terms of both major oxides and trace elements, particularly the high fluctuations in the base metals and mobile elements. These data indicate that more than one lithological unit, alteration processes and/or different fluid systems were simultaneously effective in the study area.

4.5.2. Principal Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA)

PCA was applied to the entire geochemical dataset collected from the Durmuştepe license area (n = 21; including ore and host-rock/alteration samples) in order to determine the distribution behavior of the elements and the relationships between the lithogenic background and the mineralization [63,66,91,92,93]. The first principal component (PC1) explains 52.70% of the total variance obtained from the analysis, and the second principal component (PC2) explains 17.04%. The biplot prepared for the PC1–PC2 plane shows that the sample groups and the element vectors are well separated from one another (Figure 6).

PC1 separates the mineralization from the host-rock/detrital input. On the positive side are the ore samples and Fe₂O₃, MnO, CaO, P₂O₅, trace metals and the REEs; on the negative side are the host-rock samples and SiO₂, Al₂O₃, TiO₂, K₂O, Na₂O and Ba. The positive side reflects oxy-hydroxide precipitation and the negative side detrital dilution.

On the positive part of PC2 are stable elements such as Zr, Hf and Nb, together with Ce. The decoupling of Ce from the other REEs and its concentration in this field indicate Ce fractionation developed under the redox conditions of the environment. On the negative side are MgO, Cr₂O₃, Ni, Co and the serpentinite samples (‘M1G’, ‘H5’). This distribution reflects the geochemical influence of the ophiolitic/ultramafic host rocks in the Maden Complex. While PC1 controls the distinction between mineralization and host rock, PC2 indicates the influence of the mafic–ultramafic rocks and the redox/stable-element (Ce, Zr, Hf, Nb) fractionation.

Serpentinite samples M1G and H5 record the highest absolute Ni concentrations in the dataset (2639 and 2829 ppm, respectively; Table 2), reflecting the ophiolitic ultramafic component of the Maden Complex host succession. In PCA and HCA, however, these samples cluster with other host rocks and are clearly separated from the Fe–Mn ore group on PC1. Ore samples show comparable to slightly higher Ni/Co ratios (23–31) than M1G (~19) despite similar absolute Ni levels, consistent with preferential Ni (and Co) scavenging during hydrothermal Fe–Mn oxide precipitation rather than simple mechanical mixing with serpentinite. We therefore interpret ore Ni–Co enrichment as an exhalative-hydrothermal overprint superimposed on, but not dominated by, local ultramafic input.

Hierarchical Cluster Analysis (HCA) was applied to classify the sample and element groups [66,91,94]. The clustering was carried out in two directions, for the samples and for the elements (Figure 7).

The sample dendrogram (Figure 7A) is divided into two main groups. The first group (orange) comprises the ‘Fe-‘-coded ore samples, and the second group (green) the host-rock and alteration samples (M1G, H5, Ma-1/2, D3, H9–H13, Fe-9a, G1). This bimodal distribution is consistent with the PC1 contrast in the PCA and indicates that the mineralization and the host rocks are geochemically of a completely different character.

The element dendrogram (Figure 7B) is divided into three main groups. The first main group (orange) comprises Fe₂O₃, MnO and the rare earth elements (REEs) other than Ce, together with Y, As, W, V, CaO, Mo, Pb and P₂O₅. This association may reflect elements that co-vary with Fe–Mn oxy-hydroxide precipitation; on its own it does not demonstrate contemporaneous precipitation. In the second group (green), the stable elements (Zr-Hf-Nb) are clustered with Ce. The decoupling of Ce from the other REEs is consistent with cerium fractionation. The other element groups (Al₂O₃, TiO₂, Na₂O, Ba, K₂O, TOT/C and MgO-Cr₂O₃-Zn) separate from these clusters with low similarity and represent the lithogenic and mafic–ultramafic components entering the system.

4.5.3. Correlation Matrix, Regression and Factor Analysis

Correlation analysis was carried out to define the relationships and paragenetic associations among the elements. Because the dataset did not meet the parametric assumptions and contained outliers, non-parametric methods were preferred [63].

The Shapiro–Wilk test determined that the data were not normally distributed (p < 0.05), and in order to avoid distributional [64,95] and mathematical biases, the element relationships were evaluated through the Kendall correlation matrix (Figure 8; [94]). The correlation coefficients obtained indicate that the elements precipitated contemporaneously under similar Eh–pH conditions or underwent isomorphic substitution in the crystal lattices of the minerals [94,96].

In the correlation matrix, a positive relationship (τ = 0.49) was found between Fe₂O₃ and MnO. As (MnO, τ = 0.53), Y (Fe₂O₃, τ = 0.67), Co (Fe₂O₃, τ = 0.64), as well as La (τ = 0.82) and Nd (τ = 0.71), also show positive covariation with the Fe–Mn components; it may be considered that the trace elements and REEs were enriched together with these phases. The relationships between Cr (τ = 0.16) and Ni (τ = 0.42) are weaker than those of the Fe–Mn components. While CaO shows no clear relationship with Fe₂O₃ (τ = −0.16) or MnO (τ = 0.05), it gives a positive relationship with TOT/C (τ = 0.48), which suggests that the Ca enrichment may be related to a later process, distinct from the primary Fe–Mn mineralization.

Binary scatter diagrams and regression analyses were plotted to determine the geochemical origin and enrichment trends (Figure 9).

The scatter diagrams and the regression parameters (R²) indicate that the geochemical structure is controlled by two opposing processes—detrital dilution and element scavenging:

In the regression diagrams, the negative relationship of SiO₂ and Al₂O₃ with the Fe and Mn oxides (R² = 0.53–0.93) indicates a binary mixing (dilution) model. The strong inverse relationship of Al₂O₃ with MnO (R² = 0.93) shows that terrestrial detrital material reduced the primary oxy-hydroxide accumulation [97].

The positive relationships of Lu (R² = 0.50) and As (R² = 0.44) with MnO support the retention of REEs and arsenic compounds by adsorption on manganese-oxide surfaces [86,87,98]. The strong positive relationship of Co (R² = 0.76) and Cu (R² = 0.78) with Fe₂O₃ indicates that iron oxy-hydroxides may be one of the likely carrier phases. The Co enrichment is explained by the oxidation of Co²⁺ ions to the Co³⁺ form on the iron-oxide surface and their incorporation into the lattice structure [81,99].

Factor analysis was applied as an exploratory tool to examine multivariate geochemical relationships among the ore samples (n = 11). Although the Kaiser criterion suggested six factors, scree-plot inspection and parsimony for the available sample size led to retention of four factors (Figure 10A,B;

Table 5. Varimax-rotated factor loadings and communalities for the ore samples (n = 11; 44 geochemical variables).

Variable	F1	F2	F3	F4	Communality
SiO2	0.883	−0.556	0.050	0.027	1.091
Al2O3	0.672	−0.752	−0.020	0.168	1.045
Fe2O3	−0.984	0.118	−0.102	−0.298	1.081
MgO	0.501	−0.908	0.060	−0.026	1.081
CaO	0.302	0.834	0.376	0.308	1.024
Na2O	0.665	−0.785	0.031	−0.084	1.067
K2O	0.037	0.164	−0.190	0.967	0.999
TiO2	0.634	−0.819	0.051	−0.083	1.082
P2O5	−0.239	0.276	0.661	−0.395	0.727
MnO	−0.734	0.734	0.015	0.021	1.077
Cr2O3	−0.304	0.255	−0.066	0.879	0.934
LOI	−0.547	0.317	−0.736	0.024	0.942
TOT_C	0.316	0.310	0.767	0.269	0.856
TOT_S	0.663	−0.785	0.034	−0.094	1.066
Au	0.265	0.460	−0.610	0.492	0.895
Co	−0.940	0.313	−0.122	−0.018	0.997
Hf	0.436	−0.939	0.014	−0.102	1.081
Nb	0.528	−0.893	0.016	−0.100	1.087
V	0.002	0.836	−0.186	0.436	0.924
W	−0.451	0.452	−0.604	0.064	0.777
Zr	0.153	−1.020	−0.021	−0.142	1.085
Y	−1.015	0.141	0.020	0.091	1.059
Mo	0.142	0.666	0.418	0.457	0.847
Cu	−0.857	0.059	−0.148	−0.391	0.912
Pb	−0.051	0.813	−0.064	0.263	0.737
Zn	−0.608	0.632	0.018	0.032	0.770
Ni	−0.760	0.662	−0.073	0.111	1.032
Ba	0.343	−0.544	0.302	0.379	0.648
As	−0.219	0.736	0.544	−0.271	0.959
Hg	0.166	0.554	0.183	0.753	0.935
La	−0.940	0.443	−0.041	−0.072	1.086
Ce	0.219	−0.993	0.035	−0.077	1.041
Pr	−0.983	0.318	−0.087	−0.044	1.077
Nd	−1.001	0.263	−0.112	−0.021	1.085
Sm	−1.009	0.219	−0.088	0.038	1.075
Eu	−0.985	0.276	0.026	0.078	1.053
Gd	−0.993	0.302	0.004	0.063	1.081
Tb	−1.025	0.185	−0.071	0.001	1.091
Dy	−1.010	0.246	0.004	0.011	1.081
Ho	−1.035	0.148	−0.038	−0.001	1.095
Er	−1.031	0.143	−0.081	−0.042	1.092
Tm	−1.042	0.019	−0.081	−0.008	1.093
Yb	−1.037	0.043	−0.086	−0.021	1.085
Lu	−1.026	0.039	−0.099	−0.079	1.070

). The Varimax-rotated factor loadings and communalities are reported in Table 5.

Factor 1: Factor 1 is characterized by strong negative loadings of Fe₂O₃, MnO, Co, Cu and the REE suite (La–Lu, except Ce), opposed by positive loadings of SiO₂, Al₂O₃, Na₂O, TiO₂, MgO and TOT/S. This may reflect the geochemical contrast between siliceous–detrital components and the Fe–Mn–REE ore signature during Fe–Mn oxy-hydroxide accumulation [100] rather than a unique genetic end-member.

Factor 2: Factor 2 shows positive loadings of CaO, MnO, V, Pb, Zn, Ni, Mo and As, and negative loadings of MgO, Na₂O, TiO₂, Zr, Hf, Nb and Ce. This association may reflect base-metal enrichment and its inverse relationship with immobile HFSE-bearing detrital phases, consistent with Ce fractionation under oxic seawater [101] in the ore subset.

Factor 3: Factor 3 is defined by positive loadings of TOT/C, P₂O₅ and As, with a negative LOI loading. This grouping may reflect carbon- and phosphorus-related geochemical variability among the ore samples.

Factor 4: Factor 4 is marked by positive loadings of K₂O, Cr₂O₃ and Hg, with subordinate Au and V loadings. These loadings may reflect potassium–chromium–mercury associations; they are interpreted cautiously because of the small sample size; W retention in Fe–Mn oxy-hydroxides is documented for active hydrothermal systems [102].

4.5.4. Compositional Data Analysis (CoDA)

Because of the constant-sum constraint, the geochemical data were treated as compositional data [64]. The data were therefore subjected to a CLR transformation and CoDA-PCA was performed [65].

The element relationships were examined using the log-ratio variance matrix (

Table 6. Selected low- and high-variance log-ratio element pairs for major oxides, trace elements and REEs in the Durmuştepe Fe–Mn dataset (five lowest- and five highest-variance pairs per data block).

Data Block	Variance Type	Rank	Component 1	Component 2	Log-Ratio Variance
Major oxides	Low variance	1	SiO2	Al2O3	0.1271
Major oxides	Low variance	2	Fe2O3	P2O5	0.1281
Major oxides	Low variance	3	MgO	TiO2	0.1988
Major oxides	Low variance	4	Fe2O3	Cr2O3	0.2366
Major oxides	Low variance	5	SiO2	P2O5	0.2609
Major oxides	High variance	1	Na2O	MnO	7.2660
Major oxides	High variance	2	MgO	MnO	4.6008
Major oxides	High variance	3	TiO2	MnO	4.5350
Major oxides	High variance	4	CaO	Na2O	3.4626
Major oxides	High variance	5	Na2O	Cr2O3	3.3147
Trace elements	Low variance	1	Co	Y	0.0125
Trace elements	Low variance	2	Hf	Zr	0.0308
Trace elements	Low variance	3	Co	Zn	0.0424
Trace elements	Low variance	4	Hf	Nb	0.0534
Trace elements	Low variance	5	Co	Cu	0.0637
Trace elements	High variance	1	Pb	Ba	5.0223
Trace elements	High variance	2	Nb	Pb	4.8444
Trace elements	High variance	3	Ba	As	4.6160
Trace elements	High variance	4	Nb	As	4.0469
Trace elements	High variance	5	Hf	Pb	3.9760
REEs	Low variance	1	Tm	Yb	0.0004
REEs	Low variance	2	Pr	Nd	0.0006
REEs	Low variance	3	Ho	Er	0.0011
REEs	Low variance	4	Yb	Lu	0.0012
REEs	Low variance	5	Eu	Dy	0.0013
REEs	High variance	1	La	Ce	0.4574
REEs	High variance	2	Ce	Nd	0.3695
REEs	High variance	3	Ce	Pr	0.3678
REEs	High variance	4	Ce	Sm	0.3009
REEs	High variance	5	Ce	Gd	0.2791

; Figure 11). Low variance values suggest that the elements co-vary, whereas high values suggest the influence of different geochemical processes [64].

In the REE group, the highest log-ratio variances are between Ce and the other lanthanides (La-Ce: 0.457; Ce-Nd: 0.369). Ce varies independently relative to the other lanthanides [64].

The CoDA-PCA biplot and loading plots prepared with the CLR-transformed data (Figure 12 and Figure 13) display the geological processes independently of the closure effect [64,65].

The compositional data analysis (CoDA) results re-evaluated the element relationships in the Durmuştepe dataset while overcoming the closure effect, providing a more reliable interpretation of the geochemical processes. The implications of these results are discussed below in three distinct categories:

Major Oxides: In the log-ratio variance matrix (Table 6; Figure 11), the lowest variances among the major oxides are between SiO₂ and Al₂O₃ (0.1271), Fe₂O₃ and P₂O₅ (0.1281), and MgO and TiO₂ (0.1988). These low variances reflect strong coherence and a shared geological origin. SiO₂, Al₂O₃, and TiO₂ represent the detrital clay mineral input from the host-rock sequence, whereas the Fe₂O₃–P₂O₅ pair reflects the hydrothermal precipitation of iron oxides and their associated phosphate adsorption. Conversely, the highest variances are associated with MnO and Na₂O (e.g., Na₂O–MnO: 7.2660; MgO–MnO: 4.6008), demonstrating that manganese mineralization is geochemically decoupled from both the detrital clay input and primary basaltic/ophiolitic rock signatures. This decoupling is also clear in the CoDA-PCA biplot (Figure 12), where MnO projects in a direction nearly orthogonal to SiO₂ and Al₂O₃, confirming that manganese deposition occurred as a distinct, independent hydrothermal-exhalative phase rather than a detrital constituent.

Trace Elements: For the trace elements, the lowest log-ratio variances are observed between Co and Y (0.0125), Hf and Zr (0.0308), and Co and Zn (0.0424). The close association of Hf–Zr is typical of zircon and stable detrital heavy minerals. The extremely low variance of Co with Y and Zn suggests that these metals co-varied during the primary accumulation of the iron-manganese oxides, where they were rapidly co-precipitated or scavenged from seawater. In contrast, the highest log-ratio variances occur between Pb and Ba (5.0223), Nb and Pb (4.8444), and Ba and As (4.6160). These high variances indicate different geochemical pathways. Lead (Pb) and barium (Ba) are highly mobile in hydrothermal fluids but tend to precipitate under different chemical barriers (such as sulfate vs. carbonate/oxide saturation), leading to local variations. The high variance of arsenic (As) relative to barium and niobium indicates that its enrichment was controlled by adsorption onto amorphous iron-manganese oxyhydroxides close to the hydrothermal vents, which is highly localized compared to the more regional dispersion of other trace elements.

Rare Earth Elements (REEs): In the REE subcomposition, the most significant geochemical finding is the high log-ratio variance between Ce and all other lanthanides (e.g., La–Ce: 0.4574; Ce–Nd: 0.3695; Ce–Pr: 0.3678), whereas the other REE pairs exhibit extremely low variances (e.g., Tm–Yb: 0.0004; Pr–Nd: 0.0006). This indicates that Ce behaves independently from the rest of the rare earth elements. While the trivalent REEs remained tightly coherent during transport and precipitation, Ce was subject to redox-controlled fractionation. In the CoDA-PCA biplot (Figure 12), the Ce loading projects in the opposite direction to the other REEs, mirroring the negative Ce anomaly observed in the PAAS-normalized patterns. This independent behavior of Ce is driven by the rapid oxidation of soluble Ce³⁺ to insoluble Ce⁴⁺ in well-oxygenated bottom waters and its selective adsorption onto manganese oxides, whereas the other REEs reflect the primary hydrothermal source and seawater mixing. Because these multivariate and compositional interpretations are based on a dataset of only 11 ore samples, they must be evaluated within the context of this sample size limitation. The number of geochemical variables exceeds the number of observations; consequently, the multivariate models are exploratory and may be over-parameterized. Genetic inferences in this study rely primarily on integrated field observations, ore mineralogy, trace-element and REE discrimination diagrams, and regional geological analogues; the statistical analyses provide complementary, qualitative support rather than independent confirmation.

5. Discussion

Numerous manganese deposits that developed in different geological settings occur throughout Türkiye. On the basis of their host-rock characteristics and stratigraphic ages, these deposits are classified into four main groups: those associated with radiolarian cherts, those related to black shale series, those developed in volcanic-arc settings, and those within Oligocene sedimentary units [103,104].

Based on field observations and mineralogical data, the Durmuştepe Fe–Mn mineralization falls within the group of manganese deposits that occur as lenses within radiolarian cherts. Similarly, the Vezirler oceanic Mn deposit in the Kula (Manisa) region [75], the Büyükmahal Mn occurrence in the Yozgat Artova ophiolitic complex [76] and the Kumluca Mn deposit within the Antalya Complex [77]—observed as interbeds with, or lenses within, radiolarian cherts in ophiolitic sequences and tectonic mélanges—are typical representatives of this group. These deposits are reported to have developed as a result of the mixing of hydrothermal fluids with seawater. By contrast, the Permian Zunyi [78] and Early Cambrian Maowanli [79] manganese carbonate (rhodochrosite, kutnahorite) deposits, located on the margin of the Yangtze Platform in southwestern China, can be cited as examples of deposits associated with black shale series. In contrast to the oxidizing conditions at Durmuştepe, these deposits are characterized by carbonate mineralogy that formed in anoxic–suboxic deep basins within organic-matter-rich black shales.

Compared with global deposits, the Durmuştepe Fe–Mn mineralization has a composition that is relatively poor in MnO but rich in Fe₂O₃. Whereas the Fe/Mn ratio is generally around 1 in hydrogenetic deposits that precipitate at rates of millimeters per million years in open-marine settings, this ratio can be below 0.1 or above 10 in seafloor hydrothermal systems [105,106].

The mean Fe/Mn ratio of the ore samples is 1.62 (Table 1); however, sample Fe-11 records a higher value (2.28), whereas the other ore samples for which Fe/Mn is reported range from 1.30 to 1.89 (median 1.50). We interpret the Fe-11 value as local variability within the mineralized horizon rather than as evidence of a separate genetic population. The Fe/Mn signature, together with plotting of the samples in the ‘Hydrothermal/Exhalative’ field on trace-element ternary diagrams (e.g., the Fe–Mn–(Ni + Co + Cu) × 10 diagram; Figure 3B), indicates formation near a proximal vent. Similar geochemical characteristics have been demonstrated in the Tokoro (Japan) hydrothermal Mn-oxide deposit [107], the Kasımağa (Kırıkkale) Mn-oxide deposits [108] and the Dongping sedimentary Mn deposit in the South China Block [80].

In the Kula–Vezirler radiolarite-related Mn deposit, on the other hand, the Mn/Fe ratio of higher than 5 (mean 95.26) may reflect a depositional environment distal from the feeder vent [75]. An average Fe/Mn ratio of 1.62 and co-precipitation of both metals indicate that metal-bearing fluids accumulated near the vents at the seafloor under abrupt physicochemical changes.

The petrographic examinations of the Durmuştepe Fe–Mn mineralization showed that the ore occurs within silicified mudstones and has a genetic relationship with the mafic volcanic units of the Maden Complex. The high SiO₂ contents of the samples are a result of the intense silicification processes marked by secondary quartz and chalcedony veinlets [109]. The high Al₂O₃ values in the analyses, on the other hand, are directly related to the terrestrial detrital components derived from the mudstones that constitute the host rock. This is consistent with the wide distribution limits in the box plots and shows that detrital-material input also reached the basin contemporaneously with the hydrothermal processes.

Whereas in typical sedimentary deposits the average Al value is 8.82% and Ti is 0.91%, in the Durmuştepe ores Al₂O₃ was measured as 4.22% and TiO₂ as 0.14%. While the immobile behavior of Ti within hydrothermal solutions indicates terrestrial detrital input [110], Al points to the presence of clay minerals. In Factor 1, Ti, Al and K load negatively, consistent with physical dilution of the primary ore by detrital material transported into the basin together with hydrothermal exhalations (dilution effect) [88].

Fe and Mn co-vary in the Kendall rank correlation matrix and cluster together in CoDA-PCA (PC1) and Factor 1, supporting their linked behavior in the ore-forming system [26,111,112].

The absence of a distinct positive Eu anomaly in the samples (Eu/Eu* ≈ 1.06–1.18; Figure 4) indicates that the temperature of the fluid reaching the seafloor remained below 250–300 °C. This pattern is consistent with hydrothermal fluids leaving feeder fractures and undergoing rapid, high-volume mixing with cold, oxic seawater at the seafloor [86,87,88]. Abrupt seawater contact likely drove co-precipitation of Fe and Mn. The presence of pentlandite [(Ni,Fe)₉S₈] detected in the gabbro samples can also be interpreted as one of the mineralogical indicators of the ophiolitic rock–fluid interaction that took place at the feeder base of the hydrothermal system.

In the literature, sedimentary formations with a Ce anomaly coefficient ranging between 0.2 and 0.5 are classified as deep-sea cherts [113,114]. The Ce/Ce* values of the Durmuştepe ore samples (0.10–0.15; mean 0.11) fall below this range. We interpret the unusually low Ce/Ce* chiefly as evidence of intense oxic seawater interaction and hydrothermal overprint during Fe–Mn oxide precipitation at a proximal seafloor exhalative setting within the Maden marginal basin [1,74,75,115], rather than as a direct palaeobathymetric depth estimate. The negative Ce anomaly is consistent with the decoupling of Ce from the other lanthanides in the CoDA results (La-Ce log-ratio variance: 0.457) and with oxic-seawater influence on Ce behavior [74,101,115] (Figure 4; see also the Figure 3D discussion in Section 4.3). The integrated field, mineralogical and multi-diagram evidence supports a volcano-sedimentary hydrothermal-exhalative model (Section 5) rather than a deep-basin hydrogenetic or anoxic carbonate setting [78,79]. Palaeobathymetric reconstruction from Ce/Ce* alone is not attempted here.

Compared with reference formations worldwide and in Türkiye, the Durmuştepe Fe–Mn mineralization occupies a distinct geochemical position. Durmuştepe contains much higher silica (SiO₂ = 38.24%) and aluminum (Al₂O₃ = 8.58%) than Indian Ocean and Southwest Pacific hydrogenetic Fe–Mn crusts [82,115]. Its strong negative cerium anomaly (Ce/Ce* = 0.11) contrasts with these crusts (total REEs = 2000–3800 ppm; Ce/Ce* = 1.30–2.50) and excludes a slowly precipitating hydrogenetic origin [115]. Unlike the Zunyi [78] and Maowanli [79] manganese carbonate deposits formed in anoxic–suboxic basins, Durmuştepe records a deep negative Ce anomaly (0.11), indicating a strong oxic seawater influence in the depositional environment. The ophiolitic-chert-related Kula Vezirler [75], Aşağı Eğerci [85] and Büyükmahal [76] deposits in Türkiye (with Ce anomalies of 0.58, 0.58 and 0.50, respectively), as well as the high-silica Kumluca [77] deposit with its strong positive Ce anomaly (2.34), also differ from Durmuştepe. The distinctive position of Durmuştepe among these deposits (Fe/Mn = 1.62; Ce/Ce* = 0.11; Eu/Eu* = 1.13) supports the proximal volcano-sedimentary hydrothermal-exhalative genetic model.

Figure 14 integrates the geological, mineralogical and multivariate geochemical evidence into a proximal volcano-sedimentary hydrothermal-exhalative model for the geodynamic development and formation of the Durmuştepe Fe–Mn mineralization. These hydrothermal processes can be related to geothermal circulation triggered by magmatism during the evolution of the Middle Eocene Maden marginal basin and to the leaching of the underlying spilitic basalts/diabases [39].

6. Conclusions

Based on field, petrographic, geochemical and multivariate statistical (CoDA, PCA, factor analysis) work on the Durmuştepe Fe–Mn mineralization within the siliceous mudstones of the Middle Eocene Maden Complex, the genetic model is outlined below along three main lines of evidence:

The average Fe/Mn ratio of 1.62 in the Durmuştepe Fe–Mn mineralization and the concentration of high iron contents around the vents at the gabbro–diabase contacts indicate that the metal-bearing solutions precipitated rapidly upon reaching the seafloor. The clustering of Fe and Mn in the same group in the PCA and HCA models suggests that the fluids co-varied near the feeder vents without undergoing geochemical fractionation. On the other hand, the high Ni contents, the very low Co/Ni ratios (0.03–0.04) and the hydrothermal clustering on the trace-element discrimination diagrams (Figure 3A–C) support the exhalative origin of the mineralization. The negative Ce anomaly in the PAAS-normalized patterns and the absence of a distinct positive Eu anomaly (Figure 4) indicate that vent fluids did not exceed ~250–300 °C and underwent rapid mixing with cold, oxic seawater at the seafloor (Figure 4; see also the redox diagram discussion in Section 4.3). Finally, the occurrence of the mineralization as lenses within siliceous mudstones and the grouping of the detrital lithogenic elements (Ti, Al, K) under negative loadings in the factor analysis models reflect that the continuous terrestrial detrital input into the basin physically reduced the primary ore accumulation (dilution effect). The hydrothermal circulation is thought to have been triggered by magmatism during the evolution of the Maden marginal basin [39].

Taken together, the geochemical data and multivariate models (CoDA, PCA, factor analysis) indicate that the Durmuştepe deposit formed as a proximal volcano-sedimentary hydrothermal-exhalative system, shaped by rapid mixing of low- to medium-temperature hydrothermal fluids from the ophiolitic basement of the Maden Complex with oxic seawater and by contemporaneous sedimentation [1,39]. In addition to drilling-based exploration to be carried out in the field, the joint evaluation of multi-source geological data and machine-learning-based methods may be useful for determining the mineralization potential [116].