Diagenetic Pore Fluid Evolution and Dolomitization of the Silurian and Devonian Carbonates, Huron Domain of Southwestern Ontario: Petrographic, Geochemical and Fluid Inclusion Evidence

Core samples from two deep boreholes were analyzed for petrographic, stable and Sr isotopes, fluid inclusion microthermometry and major, minor, trace and rare-earth elements (REE) of different types of dolomite in the Silurian and Devonian carbonates of the eastern side of the Michigan Basin provided useful insights into the nature of dolomitization, and the evolution of diagenetic pore fluids in this part of the basin. Petrographic features show that both age groups are characterized by the presence of a pervasive replacive fine-crystalline (<50 µm) dolomite matrix (RD1) and pervasive and selective replacive medium crystalline (>50–100 µm) dolomite matrix (RD2 and RD3, respectively). In addition to these types, a coarse crystalline (>500 µm) saddle dolomite cement (SD) filling fractures and vugs is observed only in the Silurian rocks. Results from geochemical and fluid inclusion analyses indicate that the diagenesis of Silurian and Devonian formations show variations in terms of the evolution of the diagenetic fluid composition. These fluid systems are: (1) a diagenetic fluid system that affected Silurian carbonates and was altered by salt dissolution post-Silurian time. These carbonates show a negative shift in δ18O values (dolomite δ18O average: −6.72‰ VPDB), Sr isotopic composition slightly more radiogenic than coeval seawater (0.7078–0.7087), high temperatures (RD2 and SD dolomite Th average: 110 °C) and hypersaline signature (RD2 and SD dolomite average salinity: 26.8 wt.% NaCl eq.); and (2) a diagenetic fluid system that affected Devonian carbonates, possibly occurred during the Alleghenian orogeny in Carboniferous time and characterized by a less pronounced negative shift in δ18O values (dolomite δ18O average: −5.74‰ VPDB), Sr isotopic composition in range with the postulated values for coeval seawater (0.7078–0.7080), lower temperatures (RD2 dolomite Th average: 83 °C) and less saline signature (RD2 dolomite average salinity: 20.8 wt.% NaCl eq.).


Introduction
The Michigan Basin has been the subject of several diagenetic studies in the last three decades (e.g., [1][2][3][4][5][6][7][8]). However, very few comprehensive investigations regarding the nature and composition of diagenetic fluids, lateral extent of the diagenetic processes and paragenetic sequence of different types of dolomite within Silurian and Devonian successions in the Huron Domain have been carried out. Most previous studies focused on the Paleozoic carbonate succession in the area were limited in stratigraphic and/or spatial resolution (e.g., [1][2][3][4]). In addition, contrasting models have been proposed to explain the fluid flow evolution and migration in the Michigan Basin, such as migration of dolomitizing fluids from the center of the Michigan Basin toward the margins [5] or from the back reef downward through  [16] and OPG [17]); (b) generalized basement structural contours (meters above the sea level datum) and location of structural arches and basins (modified from OPG [17] and Johnson et al. [18]).
The rifting that separated North America and Africa occurred during Precambrian to Cambrian represents the initial episode of subsidence (approximately 1100 Ma) and deposition within the Michigan Basin [10]. Thermal subsidence of the Precambrian basement followed this event approximately 580 Ma to 500 Ma because the cooled lithosphere becomes denser [19]. The subsequent and continuous deposition of sediments into the Michigan Basin caused the bending of the basement [20][21][22]. However, the development of the basin is not only the result of continuous subsidence but represents the result of a series of tectonic events that characterized the Paleozoic [23,24].
The Taconic and Acadian orogenies were a dominant control on the Paleozoic strata of the eastern flank of the Michigan Basin [25]. In fact, during the Taconic Orogeny (Ordovician) a largescale eastward-tilting of the Laurentian margin governed the disappearance of the concentric depositional geometry of the Michigan Basin with the consequent deposition of Ordovician limestone and overlying shale successions [24,26,27]. The regional maximum principal stress at that time was northwesterly oriented the same as the main direction of the thrust motion along the Appalachian tectonic front [25,28].
Restricted marine conditions occurred during the late Silurian with the consequent deposition of evaporites of the Salina Group [15,18]. In addition, the presence of an unconformity reveals emergent conditions at the end of the Silurian [25]. The Middle Devonian Acadian Orogeny reactivated the Algonquin-Findlay Arch system with the return of the concentric geometry of the Michigan Basin [23]. The Lower and Middle Devonian units are dominated by limestones and dolostones, whereas, the Upper Devonian strata are mostly represented by shales [15].  [16] and OPG [17]); (b) generalized basement structural contours (meters above the sea level datum) and location of structural arches and basins (modified from OPG [17] and Johnson et al. [18]). On a regional scale, southern Ontario is located between two major Paleozoic sedimentary basins, the Appalachian Basin to the southeast and Michigan Basin to the west ( Figure 1). The Algonquin Arch, which separates the two basins, extends from northeast of the Canadian shield to the southwest near the city of Chatham. Near the Windsor area, another structural high named the Findlay Arch starts extending to the southwest into Michigan and Ohio states [3].
The rifting that separated North America and Africa occurred during Precambrian to Cambrian represents the initial episode of subsidence (approximately 1100 Ma) and deposition within the Michigan Basin [10]. Thermal subsidence of the Precambrian basement followed this event approximately 580 Ma to 500 Ma because the cooled lithosphere becomes denser [19]. The subsequent and continuous deposition of sediments into the Michigan Basin caused the bending of the basement [20][21][22]. However, the development of the basin is not only the result of continuous subsidence but represents the result of a series of tectonic events that characterized the Paleozoic [23,24].
The Taconic and Acadian orogenies were a dominant control on the Paleozoic strata of the eastern flank of the Michigan Basin [25]. In fact, during the Taconic Orogeny (Ordovician) a large-scale eastward-tilting of the Laurentian margin governed the disappearance of the concentric depositional geometry of the Michigan Basin with the consequent deposition of Ordovician limestone and overlying shale successions [24,26,27]. The regional maximum principal stress at that time was northwesterly oriented the same as the main direction of the thrust motion along the Appalachian tectonic front [25,28].

Materials and Methods
Core samples were collected from the Bruce Nuclear site, the proposed site for a deep geological repository for low-and intermediate-level nuclear waste, from two boreholes DGR1 and DGR8 ( Figure 2) [17]. The samples were chosen based on dolomitized and fractured rocks in hand specimens. A summary of sample identification with relative age and depth is reported in Figure 3 and Table 1.
The selected core samples were photographed before making thin sections and wafers for fluid inclusions analysis. Petrographic examination of thin sections (n = 46) was performed using a standard petrographic microscope to provide lithofacies types, textural relationships and diagenetic fabric of the studied samples, to establish the paragenetic sequences. Subsequently, all thin sections were examined using cathodoluminescence microscopy (CL) to obtain data regarding the chemistry of the fluids, mineral zonation and diagenetic textures of host rock and fracture filling cement. The equipment used to carry out the cathodoluminescence analysis was a Technosyn Model 8200 MKII with a 12-15 Kv beam and a current intensity of 0.42-0.43 mA (Cambridge Image Technology Ltd., Hatfield, Hertfordshire, UK). The thin sections were also stained using a mixture of Alizarin Red S and potassium ferricyanide dissolved in a dilute solution of hydrochloric acid [35,36] to distinguish between ferroan or non-ferroan phases of calcite and dolomite. A total of forty-three (Silurian) and thirty-three (Devonian) samples were micro-sampled using a microscope-mounted drill assembly for oxygen and carbon isotopic composition from Silurian (n = 18 and n = 25) and Devonian (n = 12 and n = 21) specimens for calcitic and dolomitic components, respectively. Each sample was reacted in vacuo with 100% pure phosphoric acid four 4 h at 50 °C and 25 °C for dolomite and calcite respectively. The CO2 produced during the reaction by the different diagenetic mineral phases was extracted using the method proposed by Al-Aasm et al. [37] and measured employing a Delta-Plus mass spectrometer (Thermo Electron Corporation, Bremen, Germany). The delta values of oxygen and carbon isotopes have been reported in per mil (‰) relative A total of forty-three (Silurian) and thirty-three (Devonian) samples were micro-sampled using a microscope-mounted drill assembly for oxygen and carbon isotopic composition from Silurian (n = 18 and n = 25) and Devonian (n = 12 and n = 21) specimens for calcitic and dolomitic components, respectively. Each sample was reacted in vacuo with 100% pure phosphoric acid four 4 h at 50 • C and 25 • C for dolomite and calcite respectively. The CO 2 produced during the reaction by the different diagenetic mineral phases was extracted using the method proposed by Al-Aasm et al. [37] and measured employing a Delta-Plus mass spectrometer (Thermo Electron Corporation, Bremen, Germany). The delta values of oxygen and carbon isotopes have been reported in per mil (% ) relative to the PeeDee Belemnite (VPDB) standard and corrected for phosphoric acid fractionation. Precision was better than 0.05% for both 13 C and 18 O. Carbon dioxide extraction and isotopic composition measurements were performed at the Stable Isotopes Laboratory, University of Windsor (Windsor, ON, Canada). Around 10-50 mg of powdered samples (n = 22) were extracted to determine the Sr isotopic composition using a microscope-drill assembly, to be subsequently reacted with 2.5 N HCl for 24 h at room temperature. To avoid Strontium contamination from pore salts produced by drying the samples, they were washed twice using Milli-Q water. To measure the strontium isotope ratio in each sample, an automated Finnigan MAT 261TM mass spectrometer was employed, performing all the analysis in the static multicollector mode with Re filaments and normalizing 87 Sr/ 86 Sr ratios to 87 Sr/ 86 Sr = 8.375209. The mean standard error of the static multicollector was 0.00003 for NBS-987. These analyses were done at the University of Bochum, Bochum, Germany.
A thorough petrographic investigation of doubly polished thin sections was performed to identify and classify fluid inclusions trapped in diagenetic minerals. Measured ice-melting temperatures (Tmice) are reported as wt.% NaCl eq. using the relationship of Hall et al. [38] and Bodnar [39]. Detailed petrography was performed to determine the type of inclusions (primary or secondary/pseudosecondary) following the criteria proposed by Roedder [40]. For this study, attention was focused on primary liquid-vapor inclusion assemblages only. The microthermometric measurements were performed at the University of Windsor (Windsor, ON, Canada) employing a Linkam TH600 heating-freezing stage, coupled with the Olympus BX60. The device calibration was achieved using synthetic pure water and CO 2 inclusions as standard. Microthermometric analyses were completed on 126 primary fluid inclusions in selected carbonate samples from the Silurian (n = 85) and Devonian (n = 41) successions including pervasive replacive medium crystalline dolomite (RD2; n = 22 and n = 17 for Silurian and Devonian samples, respectively), saddle dolomite (SD; n = 31; observed in Silurian samples only) and late fracture-filling blocky calcite cement (BKC; n = 32 and n = 24 from Silurian and Devonian samples, respectively). Homogenization temperatures (T h ) and ice-melting temperatures (Tmice) were measured with a precision of ±1 • C and ±0.1 • C, respectively. Heating measurements were performed before freezing to avoid possible stretching of the inclusions during freezing [41].
Major, minor, trace and rare earth elements (REE) analyses were performed using inductively coupled plasma spectroscopy (ICP-MS) technique. Around 50-100 mg of powdered carbonate samples from different calcite and dolomite phases were digested utilizing 5% v/v acetic acid. Measurements were carried employing an Agilent 8800 inductively coupled mass spectrometer at the University of Waterloo (Waterloo, ON, Canada). All results are reported in part per million (ppm). The REEs' results obtained from ICP-MS analysis were normalized to PAAS (Post-Archean Australian Shales [42] [43]. The proportions of LREE (La-Nd) over HREE (Ho-Lu) were calculated via (La/Yb) SN ratios as proposed by Kucera et al. [44]. The proportions of MREE (Sm-Dy) was calculated using (Sm/La) SN and (Sm/Yb) SN ratios [45] over LREE and HREE, respectively.

Petrography of Silurian Formations
The diagenetic history of the Silurian formations includes processes such as dolomitization mechanical and chemical compaction, fracturing, dissolution, silicification, and calcite and evaporite cementation. The diagenetic processes above are described in Tortola [46], on which this paper is largely based.

Dolomitization
Four types of dolomite occur in the Silurian formations ( Figure 4) and are classified following the criteria proposed by Sibley and Gregg [47]. These include including three types of replacive matrix dolomite and one of dolomite cement. These are a pervasive replacive micro to fine crystalline Minerals 2020, 10, 140 9 of 52 (<50 µm) euhedral to subhedral dolomite matrix (RD1) ( Figure 4A); a pervasive replacive medium (>50-100 µm) euhedral crystalline dolomite matrix (RD2) ( Figure 4B), which represent the main constituents of the host rock; and a less abundant selective replacive medium (>50-100 µm) euhedral crystalline dolomite matrix (RD3) only observed along dissolution seams in partially dolomitized limestones, usually characterized by cloudy-dark cores and clear rims ( Figure 4C). All dolomite types are non-ferroan and show dull red to non-luminescent (RD1 and RD2) and red to bright (RD3) under the CL ( Figure 4D,E). In some cases, RD2 shows undulose extinction similar to saddle dolomite (e.g., [5,48]) and is characterized by cloudy non-ferroan cores followed by later clear and iron-rich rims ( Figure 4B,F). In addition to these types, a coarse crystalline (>500 µm), ferroan saddle dolomite cement (SD; Figure 4G) filling fractures and vugs are observed, commonly predating late calcite cement ( Figure 4E,H) and occasionally evaporite minerals ( Figure 4I). Typically, SD is fluid inclusion-rich, zoned with cloudy cores and clear rims, and characterized by curved crystal boundaries. ( Figure 4B,F). In addition to these types, a coarse crystalline (>500 μm), ferroan saddle dolomite cement (SD; Figure 4G) filling fractures and vugs are observed, commonly predating late calcite cement ( Figure 4E,H) and occasionally evaporite minerals ( Figure 4I). Typically, SD is fluid inclusionrich, zoned with cloudy cores and clear rims, and characterized by curved crystal boundaries.    Three types of fractures are present in the studied formations. Thin randomly oriented hairline fractures (<0.5 cm) filled with drusy calcite cement (DC) represent the first generation of fractures ( Figure 5A). Late wider subhorizontal and subvertical fractures (0.5-1 cm), both commonly occluded by saddle dolomite, late blocky calcite and evaporite cement representing the second and third generations, respectively ( Figure 5B,C). Cross-cutting relationship of the last two generations of fractures shows that cementation took place after the formation of both fractures.

Fracturing
Three types of fractures are present in the studied formations. Thin randomly oriented hairline fractures (<0.5 cm) filled with drusy calcite cement (DC) represent the first generation of fractures ( Figure 5A). Late wider subhorizontal and subvertical fractures (0.5-1 cm), both commonly occluded by saddle dolomite, late blocky calcite and evaporite cement representing the second and third generations, respectively ( Figure 5B,C). Cross-cutting relationship of the last two generations of fractures shows that cementation took place after the formation of both fractures.

Calcite Cementation
Early and late-stage calcite cement ( Figure 6) have been distinguished including from early to late: isopachous, syntaxial overgrowth, drusy, and blocky calcite. Undolomitized calcite matrix (MC) from precursor limestone was also recognized ( Figures 4C and 5A). Isopachous calcite cement (ISC) is characterized by cement rims growing around coated grains. It exhibits microcrystalline, bladed

Calcite Cementation
Early and late-stage calcite cement ( Figure 6) have been distinguished including from early to late: isopachous, syntaxial overgrowth, drusy, and blocky calcite. Undolomitized calcite matrix (MC) from precursor limestone was also recognized ( Figures 4C and 5A). Isopachous calcite cement (ISC) is characterized by cement rims growing around coated grains. It exhibits microcrystalline, bladed texture ( Figure 6A) with non-ferroan, dully-red luminescence ( Figure 4D) and crystal size ranging from 50 to 100 µm. Syntaxial calcite overgrowth cement (SXC) forms ferroan, bright-luminescent crystals ranging in size from 100 µm up to >500 µm, usually around echinoderm fragments ( Figure 6B). Void-filling and pore-lining drusy calcite cement (DC) was mainly observed in intergranular and intraskeletal pores, molds and fractures. Common textural features show equant to elongate, anhedral to subhedral crystals ranging in size between 75-250 µm with dimension increasing towards the center of the pore space ( Figure 6C). In most cases, DC is ferroan and exhibits dully-red luminescence under CL ( Figure 6D).
A late fractures and void-filling blocky calcite cement (BKC) consisting of coarse non-ferroan to ferroan crystal ( Figure 6E) ranging in size from 200 µm to >500 µm, shows zoned red to bright luminescence under CL ( Figure 6F) and commonly postdate saddle dolomite cement.
In terms of paragenesis, early calcite cement (ISC, SXC and DC) predated RD1, RD2 and RD3 and predated fracture-and pore-filling saddle dolomite which is postdated by late blocky calcite (BKC) and evaporite (GY and ANH) cement.

Evaporite Cementation
Pore and fracture-filling fibrous gypsum (GY)/anhydrite (ANH) cements in the Silurian formations range in size from 50-150 µm ( Figures 4I and 5C,D). They are the latest stage diagenetic mineral phases. Both gypsum and anhydrite are only present in samples from the Salina Group Formation, commonly postdating fracture-filling blocky calcite and saddle dolomite cement ( Figure 4I).

Dolomitization
Three types of dolomite occur in the Devonian formations (Figure 7), which are characterized by the presence of pervasive replacive micro to fine (<50 µm), euhedral to subhedral, crystalline dolomite matrix (RD1) ( Figure 7A,B), a pervasive replacive medium (>50-100 µm) euhedral crystalline dolomite matrix (RD2) ( Figure 7B-D), which represent the main constituents of the host dolostone, and a less abundant selective replacive medium (>50-100 µm) euhedral crystalline dolomite matrix (RD3 commonly associated with dissolution seams), and usually characterized by cloudy-dark core and clear rims ( Figure 7E). All three types of dolomite are non-ferroan and show non-luminescent to dull red (RD1 and RD2) and red to bright red (RD3) under the CL ( Figure 7B,F).
A schematic summary of the main features with relative nomenclatures of the main carbonate minerals observed and described for both age groups is presented in Table 2. crystals ranging in size from 100 μm up to >500 μm, usually around echinoderm fragments ( Figure  6B). Void-filling and pore-lining drusy calcite cement (DC) was mainly observed in intergranular and intraskeletal pores, molds and fractures. Common textural features show equant to elongate, anhedral to subhedral crystals ranging in size between 75-250 μm with dimension increasing towards the center of the pore space ( Figure 6C). In most cases, DC is ferroan and exhibits dully-red luminescence under CL ( Figure 6D).  Syntaxial calcite overgrowth cement (SXC) forms non-ferroan, non-luminescent crystals, usually around echinoderm fragments, ranging in size from 100 μm up to >500 μm ( Figure 9A).   The fractures observed in the Devonian formations can be divided in two types. The first is represented by sub-horizontal fractures (0.5-1 cm) filled by blocky calcite cement ( Figure 8A).
The second type is represented by sub-vertical fractures (0.1-0.5 cm) commonly filled by blocky calcite ( Figure 8A,B) crosscutting chemical compactional features ( Figure 8C,D). Dogtooth calcite cement (DTC), is characterized by non-ferroan, non-luminescent, sharply pointed acute crystals (50 μm) growing normal to the substrate. It is mainly observed lining intraskeletal chambers, and commonly presenting textures from elongate scalenohedral to rhombohedral ( Figure 9B). Pore-lining and void-filling drusy calcite cement (DC) is mainly observed in intergranular and intraskeletal pores, molds and fractures ( Figure 9C-E). Common textural features show equant to elongate, in some cases anhedral to subhedral crystals ranging in size between 75-250 μm with dimension increasing towards the center of the pore space ( Figure 9D). In most cases, DC is nonferroan ( Figure 9E) and shows zoned, dull-red to bright orange luminescence ( Figure 9C).
A late fracture and void-filling blocky calcite cement (BKC) consisting of the coarse-grained, non-ferroan crystal ( Figure 9E) without preferred orientation, ranges in size from 200 μm to >500 μm, shows zoned red to bright orange luminescence under CL ( Figure 9F).
Pore-lining and void-filling drusy calcite cement (DC) is mainly observed in intergranular and intraskeletal pores, molds and fractures ( Figure 9C-E). Common textural features show equant to elongate, in some cases anhedral to subhedral crystals ranging in size between 75-250 µm with dimension increasing towards the center of the pore space ( Figure 9D). In most cases, DC is non-ferroan ( Figure 9E) and shows zoned, dull-red to bright orange luminescence ( Figure 9C).
A late fracture and void-filling blocky calcite cement (BKC) consisting of the coarse-grained, non-ferroan crystal ( Figure 9E) without preferred orientation, ranges in size from 200 µm to >500 µm, shows zoned red to bright orange luminescence under CL ( Figure 9F).

Strontium Isotopes
Due to the mixed nature of the samples, only one pure sample of drusy calcite cement (DC), for each age group, were microsampled avoiding then contamination with other carbonate minerals. The Sr isotopic values of DC were 0.708154 and 0.708051 for the Silurian and Devonian samples, respectively ( Figure 12). The Silurian late calcite cement (BKC) show 87 Sr/ 86 Sr isotopic ratio values ranging from 0.708054 to 0.708383 (n = 2) and 0.707982 for Devonian BKC (n = 1). 87 Sr/ 86 Sr ratios of matrix dolomite samples (RD1, RD2 and RD3) from the Silurian (n = 7) and Devonian (n = 6) carbonate successions range from 0.708469 to 0.708919 and 0.707884 to 0.708920, respectively. Saddle dolomite sampled from Silurian formations has Sr isotopic values ranging from 0.708516 to 0.708648 (n = 2; Figure 12). Gypsum samples from Silurian formations show Sr isotopic values ranging from 0.708434 to 0.708486.  87 Sr/ 86 Sr ratios of different dolomitic and calcitic components from the two age groups. In both age groups, Sr isotopic ratios show seawater composition of their respective age as the primary source of diagenetic fluids with minor rock-water interactions as indicated by some exceptions of more radiogenic signatures in the Silurian formations. The green box represents Devonian seawater [51] and the yellow box represents Silurian seawater [51,52]. 87 Sr/ 86 Sr isotopic ratios for dolomite and calcite by age are presented in Table 3.

Fluid Inclusions Microthermometry
Fluid inclusions measurements were focused only on dolomite and blocky calcite cement. Due to the small size of crystal and/or fluid inclusions, microthermometric measurements in mineral phases such as RD1and RD3 were not achievable in both age groups. The melting (Tm) and homogenization temperatures (Th) were measured from two-phase, liquid-rich primary fluid inclusions which ranged in size from 1 µ m up to 20 µ m in diameter with variable shapes from irregular to nearly circular ( Figure 13). In some cases, measurements of the melting temperatures for fluid inclusions hosted in RD2 were not possible due to their small size in both Silurian and Devonian successions (measured n = 9 and n = 7 for Silurian and Devonian samples, respectively). Hydrocarbon fluid inclusions were not identified in the selected samples from the studied formations.  87 Sr/ 86 Sr ratios of different dolomitic and calcitic components from the two age groups. In both age groups, Sr isotopic ratios show seawater composition of their respective age as the primary source of diagenetic fluids with minor rock-water interactions as indicated by some exceptions of more radiogenic signatures in the Silurian formations. The green box represents Devonian seawater [51] and the yellow box represents Silurian seawater [51,52]. 87 Sr/ 86 Sr isotopic ratios for dolomite and calcite by age are presented in Table 3.

Fluid Inclusions Microthermometry
Fluid inclusions measurements were focused only on dolomite and blocky calcite cement. Due to the small size of crystal and/or fluid inclusions, microthermometric measurements in mineral phases such as RD1and RD3 were not achievable in both age groups. The melting (Tm) and homogenization temperatures (T h ) were measured from two-phase, liquid-rich primary fluid inclusions which ranged in size from 1 µm up to 20 µm in diameter with variable shapes from irregular to nearly circular ( Figure 13). In some cases, measurements of the melting temperatures for fluid inclusions hosted in RD2 were not possible due to their small size in both Silurian and Devonian successions (measured n = 9 and n = 7 for Silurian and Devonian samples, respectively). Hydrocarbon fluid inclusions were not identified in the selected samples from the studied formations.  The summaries of the fluid inclusion results are presented in Figures 14-16 and Table 4. Examples of the main characteristics of fluid inclusion assemblages hosted within the different calcite and dolomite phases are illustrated in Figure 13.  Table 4. Examples of the main characteristics of fluid inclusion assemblages hosted within the different calcite and dolomite phases are illustrated in Figure 13.

Major Elements and Dolomite Stoichiometry
The concentrations of the major elements, calcium and magnesium, were determined by ICP-MS analysis for dolomite and calcite phases to establish their stoichiometry. Ideal or stoichiometric dolomite would be characterized by the equal molar percentage of MgCO 3 and CaCO 3 and would be less soluble and more stable compared to non-stoichiometric dolomite [53]. In contrast, non-stoichiometric dolomite will have more Ca than Mg and because of its heterogeneous structure and composition is unstable and dissolve more rapidly.
All the dolomite types recognized petrographically were sampled and analyzed without any contamination from calcite components for both age groups with the only exception for the selective replacive dolomite (RD3) which could not be sampled without possible contamination by calcite or other dolomite phases in the Silurian specimens. With rare exceptions, all types of dolomite analyzed in this study were non-stoichiometric in both age groups (  Minor and Trace Elements: Iron, Manganese and Strontium The results of trace elements in dolomitic phases are presented in Table 5.  Table 6).
Replacive dolomite RD1 and RD2 sampled from Silurian formations show sub-parallel trends compared to the pattern of Silurian brachiopods shown by Azmy et al. [54] but with substantial differences from Devonian patterns, Silurian dolomite trends are characterized by a minor negative La anomaly and both cases of positive and negative Ce anomalies ( Figure 17A,B). Both types of dolomite RD1 and RD2 exhibit higher average ΣREE (11.88 ± 6.25 ppm, and 10.62 ± 8.92 ppm, respectively) compared with the average of Silurian brachiopods (2.21 ppm) shown by Azmy et al. [55]. Saddle dolomite (SD) and blocky calcite cement (BKC) show similar REE shale-normalized trends ( Figure 17A,B) with average ΣREE (47.06 ± 41.95 ppm and 28.83 ± 24.47 ppm, respectively) and both significantly differ from the Silurian Brachiopods REE patterns. As shown in Figure 17, SD and BKC present both negative La and Ce anomalies ( Figure 17B) as well as a slight negative Eu anomaly ( Figure 17A).
All the dolomite and late calcite samples from Devonian formations exhibit shale-normalized patterns comparable to those of modern warm water brachiopods ( Figure 18A).      Figure 18A,B), and slightly positive La and Gd anomalies ( Figure 18A). Replacive dolomite RD1, RD2 and RD3 exhibit higher and progressively increasing average ΣREE (1.14 ± 0.93 ppm, 1.66 ± 1.26 ppm and 6.41 ± 8.50 ppm, respectively) compared with the average of modern warm water brachiopods (0.25 ± 0.2 ppm) proposed by Azmy et al. [55]. Slightly different patterns, which are still comparable to the trend of modern warm water brachiopods, characterize REEs of BKC with ΣREE of 3.08 ppm which falls between those of RD1 and RD3 ( Figure 18A).

Discussion and Interpretations
The petrographic and geochemical results obtained in this study indicate that diagenesis of the Silurian and Devonian dolomitized successions have been subjected to fluids of variable origins and compositions, as described below.

Silurian
The diagenetic history of the Silurian formations includes dolomitization, mechanical and chemical compaction, fracturing, dissolution, silicification and calcite and evaporite cementation ( Figure 19). The diagenetic processes above listed are described in depth in Tortola [46], on which this paper is largely based.

Discussion and Interpretations
The petrographic and geochemical results obtained in this study indicate that diagenesis of the Silurian and Devonian dolomitized successions have been subjected to fluids of variable origins and compositions, as described below.

Silurian
The diagenetic history of the Silurian formations includes dolomitization, mechanical and chemical compaction, fracturing, dissolution, silicification and calcite and evaporite cementation ( Figure 19). The diagenetic processes above listed are described in depth in Tortola [46], on which this paper is largely based. Figure 19. Paragenesis of Silurian formations (modified from Tortola [46]).

Dolomitization
Replacive dolomite crystal size is often influenced by the texture of the precursor limestone [55]. Hence, fine-crystalline dolomite may represent the replacement of fine-grained limestone, whereas coarser replacive dolomite may either represent replacement of coarse-grained limestone or recrystallization of precursor dolomite matrix [55]. RD1 is commonly characterized by closely packed unimodal mosaics of micro to fine-crystalline dolomite which preserved the finely laminated texture of the precursor mudstone. The characteristics showed by RD1, are comparable to fine crystalline dolomite observed in Silurian formations from different areas of southwestern Ontario [5,7,8,56]. This dolomite has been interpreted to be formed from evaporative reflux of hypersaline brines in a sabkha-like setting in a near-surface or shallow burial setting, in the condition of relatively high salinity and low temperature [47,57,58]. This interpretation is also supported by a cross-cutting relationship with chemical compactional features, such as dissolution seams and stylolites, which evidently postdated RD1. However, there is no evidence of supratidal lithofacies and abundant evaporites.
As mentioned by Zheng [7], the relatively coarser size of replacive fine crystalline dolomite (20-50 µ m) compared to modern dolomites (<4 µ m) can be explained by an increase of crystal size due to early recrystallization. This early recrystallization is possibly supported by the negative shift in Figure 19. Paragenesis of Silurian formations (modified from Tortola [46]).

Dolomitization
Replacive dolomite crystal size is often influenced by the texture of the precursor limestone [55]. Hence, fine-crystalline dolomite may represent the replacement of fine-grained limestone, whereas coarser replacive dolomite may either represent replacement of coarse-grained limestone or recrystallization of precursor dolomite matrix [55]. RD1 is commonly characterized by closely packed unimodal mosaics of micro to fine-crystalline dolomite which preserved the finely laminated texture of the precursor mudstone. The characteristics showed by RD1, are comparable to fine crystalline dolomite observed in Silurian formations from different areas of southwestern Ontario [5,7,8,56]. This dolomite has been interpreted to be formed from evaporative reflux of hypersaline brines in a sabkha-like setting in a near-surface or shallow burial setting, in the condition of relatively high salinity and low temperature [47,57,58]. This interpretation is also supported by a cross-cutting relationship with chemical compactional features, such as dissolution seams and stylolites, which evidently postdated RD1. However, there is no evidence of supratidal lithofacies and abundant evaporites.
As mentioned by Zheng [7], the relatively coarser size of replacive fine crystalline dolomite (20-50 µm) compared to modern dolomites (<4 µm) can be explained by an increase of crystal size due to early recrystallization. This early recrystallization is possibly supported by the negative shift in oxygen isotopic composition discussed in the next section. Cathodoluminescence of mainly dull to dark red luminescence for RD1 suggests replacement in a reducing environment at shallow burial conditions [59].
Many studies (e.g., [62,64,65]) suggest that non-planar crystal boundaries, coarsening of crystal size and absence of zonation under CL can represent evidence of recrystallization. According to Machel [66], the most accredited definition of recrystallization is the one proposed by the Glossary of Geology [67], which states that recrystallization of a mineral determines the formation of a new mineral commonly characterized by coarser crystals and with either same or different composition. The coarsening of crystal size coupled with the lack of zonation under CL observed in RD2 may indicate recrystallization [64,68,69]. The occurrence of strictly non-planar crystal boundaries was not observed in RD2 from both Silurian and Devonian samples, possibly because early dolomite, formed from seawater can preserve unchanged geochemistry and texture after a certain depth of burial and several million years [70].
According to Machel [66], dolomite can be defined "significantly recrystallized" if it is characterized by the variation of at least one of the following parameters: (1) increase of crystal size or in non-planar crystal boundaries; (2) progressive ordering; (3) change in composition including stable and radiogenic isotopes, stoichiometry, trace and rare earth elements, fluid inclusions and zonation; (4) change in paleomagnetic properties.
The increase of crystal size, the negative shift of δ 18 O from postulated values of marine dolomite formed during Silurian and changes in trace and rare earth elements (discussed in the next sections), make RD2 "significantly recrystallized", but also "insignificantly recrystallized" with respect to the increase of non-planar crystal boundaries and progressive ordering [1,5,66]. Crystal coarsening of fabric-destructive pervasive replacive medium (50-100 µm) crystalline dolomite (RD2) may suggest recrystallization from the precursor dolomite RD1, but the presence of cloudy, non-ferroan cores and clear, ferroan rims as a common characteristic of RD2 crystals ( Figure 4F) may also suggest overgrowth on earlier dolomite [47]. Under CL, medium dolomite RD2 shows slightly more red luminescence compared to RD1, which indicates that its formation took place in slightly deeper burial conditions in a more reduced environment [59]. As the hypothesis of overgrowth of RD2 on early dolomite is not quite confirmed by CL observation, no brighter luminescent zones have been observed in correspondence of the clear rims, the interpretation of recrystallization seems more likely. Selective replacive medium crystalline dolomite (RD3) occurred in association with dissolution seams during early chemical compaction. Similar textures of patchy dolomite formed in association with chemical compactional features have been previously described (e.g., [71,72]) as dolomite formed from marine or connate waters flowed through dissolution seams which are believed to have behaved as preferential transport path for the dolomitizing fluids. Alternatively, the magnesium required for the formation of RD3 may be supplied by the remobilization of the Mg released in solution from the dissolution of high magnesium calcite [72,73]. The association with dissolution seams suggests that RD3 formed in shallow to intermediate burial realm postdating RD1 and RD2. The common presence of dark cores and clear rims as distinctive characteristics of RD3 crystals ( Figure 4C) may also suggest overgrowth on the previous dolomite phases [47]. This evidence, as mentioned for RD2, is not confirmed by cathodoluminescence which shows luminescence comparable with those of RD2 without any zonation or difference in terms of luminescence between cores and rims in RD3 crystals ( Figure 4D).
The less abundant saddle dolomite (SD) is observed as fracture and vug-filling cement occluding second and third generation of fracture, postdating RD1, RD2 and RD3 and predating blocky calcite (BKC) and evaporite cement (GY and ANH) ( Figure 4E,G-I). This petrographic evidence suggests that saddle dolomite formed during a late diagenetic event possibly in intermediate burial conditions.

Calcite Cementation
Calcite cementation represents an important diagenetic process that affected Silurian carbonate rocks in the Michigan Basin. It started in the early-stage, during marine environmental conditions, to continue during progressively deeper burial.
Isopachous calcite (ISC) was only observed in Silurian formations as cement characterized by bladed crystals surrounding coated grains ( Figure 6A) formed prior to mechanical compaction and consequently prior to SXC, DC and BKC. It is interpreted to have formed from seawater immediately after deposition in a near-surface marine environment, inhibiting and hence predating mechanical compaction. Under CL this cement is dull-red ( Figure 4D), occasionally non-luminescent, indicating precipitation in a slightly reducing environment where Fe 2+ and Mn 2+ were available to be incorporated in the crystal lattice [59].
Syntaxial overgrowth calcite (SXC) is mainly cloudy and inclusion-rich, evidence of precipitation in near-surface marine, meteoric or mixing-zone environments [74]. The presence of dissolution seams deflecting around and cross-cutting SXC indicates that its precipitation took place before chemical compaction. Under CL this cement is red to bright red indicating precipitation in a redox environment where Fe 2+ and Mn 2+ were available to be incorporated in the crystal lattice [59]. Isotopic values for this cement from both Silurian and Devonian rocks ( Figure 11) show overlapping and/or slightly depleted values relative to the postulated ratios for their respective ages; hence evidence for their formation form seawater parentage fluids.
Drusy mosaic calcite (DC), was mainly observed as vug, mould and fracture-filling cement ( Figure 6C), which can form in near-surface meteoric as well as in burial environment [75,76] but its formation is not commonly related to the marine environment [62]. Under CL it mainly shows dull-red luminescence ( Figure 6D), evidence of precipitation in a reducing condition [59]. Paragenetically, DC postdate silica and predated dissolution seams and stylolites ( Figure 5A) as well as late blocky calcite cement ( Figure 5B). Hence, its formation represents a relatively early diagenetic event predating chemical compaction.
Late diagenetic blocky calcite cement (BKC) is characterized by coarse crystals (>500 µm) and it is mainly observed as vug and fracture filling cement which possibly formed in a burial environment (e.g., [77]). Petrographically, BKC shows dull to bright orange zoned luminescence under CL indicating formation in a progressively reducing environment. Based on petrographic evidence, BKC is interpreted to be the latest phase of calcite cementation, postdating saddle dolomite cement (SD) ( Figure 4I) and predating evaporite cementation.

Devonian
The diagenetic history of Devonian successions includes dolomitization mechanical and chemical compaction, fracturing, dissolution, silicification, and calcite cementation ( Figure 20). These sequences are based on petrographic cross-cutting relationships supplemented by geochemical evidence.

Devonian
The diagenetic history of Devonian successions includes dolomitization mechanical and chemical compaction, fracturing, dissolution, silicification, and calcite cementation ( Figure 20). These sequences are based on petrographic cross-cutting relationships supplemented by geochemical evidence.  RD1 is commonly characterized by closely packed unimodal mosaics of micro to fine-crystalline dolomite ( Figure 7A) which preserved the original texture of the precursor mudstone. Fine-crystalline dolomites distinguished in Devonian formations in previous studies (e.g., [2,5,33,56]) show similar texture if compared to RD1 and are interpreted to be formed from evaporative reflux of hypersaline brines in a sabkha-like setting. Hence, the mimetic pervasive replacive micro to fine crystalline dolomite (RD1) is possibly formed in a low temperature, near-surface or shallow burial environment. However, there is no evidence of abundant evaporites commonly associated with this model. Slightly coarser size of RD1 crystals, in comparison with recent dolomites, suggest early recrystallization [7], which is supported by oxygen isotopic composition discussed in Section 5.2. Cathodoluminescence characteristics of RD1 showing mainly dull to dark red luminescence ( Figure 7B) suggest replacement in a shallow reducing environment [53]. Cross-cutting relationship with dissolution seams and stylolites suggests that RD1 formed prior to chemical compaction.
Crystal coarsening of fabric-destructive pervasive replacive medium crystalline dolomite (RD2) ( Figure 7C,D) suggests possible recrystallization of the precursor dolomite RD1 [47,[61][62][63][64]. The lack of zonation under CL coupled with the coarsening of crystal size observed in RD2 may represent additional evidence of recrystallization [64,68]. Devonian medium replacive dolomite shows slightly brighter red luminescence compared to RD1 ( Figure 7B), possible evidence of a slightly deeper burial formation in a more reducing environment.
Selective replacive medium crystalline dolomite (RD3) has similar characteristics to RD3 observed in Silurian formations. It is commonly associated with dissolution seams and stylolites, which act as preferential paths for dolomitizing fluids [72]. Hence, it is interpreted to have formed in a shallow to intermediate burial regime postdating RD1 and RD2 ( Figure 7E,F) and alongside with chemical compaction. The common presence of dark cores and clear rims as distinctive characteristics of RD3 crystals ( Figure 7E) may also indicate overgrowth on the previous dolomite phases [47]. This hypothesis is not confirmed by cathodoluminescence results which shows luminescence slightly brighter with those of RD2 but without any zonation or difference in terms of luminescence between cores and rims within the crystals ( Figure 7B,F).

Calcite Cementation
Calcite cementation represents an important process within the paragenesis of Devonian carbonate rocks in the Michigan Basin (Figure 9). It started in the early-stage, during marine environmental conditions, to proceed during progressively deeper burial stages.
Syntaxial overgrowth calcite cement (SXC) is commonly observed in bioclastic grainstone, usually as nucleation over echinoderm plates or crinoid ossicles that act as single crystals ( Figure 9A). Syntaxial overgrowth is mainly cloudy and inclusion-rich, evidence of precipitation in near-surface marine, meteoric or mixing-zone environments [74]. The presence of dissolution seams deflecting around and crosscutting SXC indicates that its precipitation took place before chemical compaction. Under CL this cement is dull-red indicating precipitation in a reducing environment where Fe 2+ and Mn 2+ were available to be incorporated in the crystal lattice [59].
Dogtooth calcite cement (DTC), which is only present in bioclastic grainstones from Devonian formations ( Figure 9B), can form in different environments. According to Reinhold [78] this cement can form: (1) in a marine-phreatic environment when dogtooth calcite occurs in intraskeletal voids; (2) in shallow burial environment when observed in stromatactis cavities succeeding radiaxial fibrous calcite cement. DTC is observed growing normally to the substrate (mainly in intraskeletal pores) with sharply pointed acute crystals. It formed early within the diagenetic history of the Devonian formations, in a near-surface or shallow burial environment, predating the formation of silica and consequently predating RD1. The presence of zoned dull to red luminescence under CL indicates possible formation in a progressively reducing environment [53]. As sampling of a pure specimen for geochemical analyses was not possible, the conclusion concerning time and environment of DTC's occurrence are based exclusively on petrographic observations. Drusy mosaic calcite (DC), mainly observed as vug, mould and fracture-filling cement ( Figure 9D,E), can form in near-surface meteoric as well as in burial environment [75,76]. Its formation is not commonly related to the marine environment [71]. CL observations reveal luminescent zones ranging from non-to orange bright luminescent ( Figure 9C) indicating fluctuation from well oxidizing environment, which inhibited the incorporation of Fe 2+ and Mn 2+ , to a slightly more reducing condition, respectively [59]. DC postdated silica and predated dissolution seams, stylolites and late blocky calcite cement ( Figure 9E). Hence, its formation is thought to represent a relatively early diagenetic event which possibly took place in a shallow burial environment, predating chemical compaction.
Late diagenetic blocky calcite cement (BKC) possibly formed in the burial environment. BKC shows dull to bright zoned luminescence under CL ( Figure 9F) indicating formation in a progressively reducing environment. Based on petrographic evidence, BKC is interpreted to be the latest phase of calcite cementation ( Figure 20) postdating fractures.

Constraints from Stable O, C and Radiogenic Sr Isotopes
The conditions of precipitation and/or recrystallization of diagenetic minerals can be predicted using their oxygen isotopic composition, which is a function of the temperature as well as the δ 18 O composition of the diagenetic fluids involved [50]. The recrystallization of dolomite, and associated chemical and textural changes, is considered a frequent process in carbonates associated with increasing burial depth [50,64,66,69,79]. Isotopic data show a general overlap between δ 18 O and δ 13 C values of replacive dolomites (RD2 and RD3) from Silurian and Devonian successions ( Figure 10). In both age groups, for these dolomite samples display similar δ 13 C compared to those of carbonates formed in equilibrium with seawater of the respective age (for Silurian: 0.5 to 2.5% VPDB [80]; and for Middle Devonian: 0.5 to 2.5% VPDB [49]) ( Figure 10). The δ 13 C values possibly reflect the carbon isotopic composition of the precursor limestones (e.g., [81]). However, many δ 13 C values in Silurian RD1 departed from the postulated values for Silurian carbonates. In contrast, in both age groups, the δ 18 O isotopic composition shows a negative shift from the postulated values for their respective age.

Silurian
Dolomite formed in equilibrium with seawater should display δ 18 O isotopic composition enriched of about 3 ± 1% (e.g., [82]). Calcitic matrix is characterized by δ 18 O values varying from −8.16 to −4.73% VPDB and hence with some exceptions, mostly falling into the range of the postulated values for Silurian marine calcite (−3 to −5.5% VPDB [80]). Assuming that the Silurian seawater with an isotopic composition of −3.5 % SMOW [83] represents the source of dolomitizing fluids for the formation of RD1, its range of δ 18 O values should vary from −0.25 to −2.5% VPDB. The calculated average value (−1.4% VPDB) would correspond to an estimated value of 24 • C employing the fractionation equation proposed by Land [82]. However, values obtained from RD1 show a negative shift of about 2-6% from the δ 18 O average value (−1.4% VPDB) expected for the Silurian dolomite ( Figure 10). RD2, RD3 and SD exhibit a negative shift of about 2-8% compared with the δ 18 O average value of the Silurian marine dolomite (Figure 10). An explanation to justify this negative shift of δ 18 O values can be the interaction with meteoric water or formation in the burial regime at relatively high temperatures (e.g., [69]). The interaction with meteoric water should be demonstrated by evidence of subaerial exposure such as karst phenomena, presence of vadose cement or other features associated with meteoric diagenesis. Although previous studies underline the presence of an erosional unconformities (between Cabot Head and the overlying Fossil Hill formations and between Bass Island and Bois Blanc formations [34]), the lack of evidence of karstification in the studied samples (e.g., [69]), make the meteoric diagenesis an improbable scenario for matrix dolomites. Hence, the hypothesis of recrystallization with burial seems to be more likely.
RD1 shows negative δ 13 C values, possibly related to the formation of this dolomite in a shallow burial environment associated with bacterial sulphate reduction (BSR) [84]. This hypothesis is supported by the presence of minute framboidal pyrite crystals ( Figure 4A) embedded within RD1 [85].
Even though, there is no evidence of the corresponding variation in terms of crystal size coarsening and depletion of δ 18 O values (e.g., [8]) comparing data from RD1 and coarser dolomites (RD2 and RD3), the negative shift of δ 18 O values could be associated to recrystallization and/or to interaction with hydrothermal fluids at relatively high temperature (e.g., [5,69,77,86,87]).
Most of the dolomite samples (both, matrix and cement) from Silurian formations are characterized by 87 Sr/ 86 Sr ratios falling in the range of postulated values for Silurian seawater (0.7078-0.7087 [52,53]) ( Figure 12). However, some samples show slightly more radiogenic values than those of coeval seawater. This variation from the postulated values can be related to the recrystallization of early dolomite and to water/rock interaction with feldspar-bearing rocks [5]. In an overall view, the 87 Sr/ 86 Sr ratios for dolomite matrix (RD1, RD2 and RD3), dolomite cement (SD) and late-stage calcite cement (BKC) suggest for both a seawater-dominated parental fluid.
Most of the calcitic samples show δ 13 C values, in range with postulated values for Silurian marine calcite with the exception for late blocky calcite which clearly shows a negative shift from the equilibrium values ( Figure 11). This may suggest a separate input of oxidized organic carbon possibly related to interaction with hydrocarbons (e.g., [77]).

Devonian
Calcitic matrix is characterized by δ 18 O values varying from −8.16 to −4.73% VPDB and hence with some exceptions (Figure 11), mostly falling into the range of the postulated values for Devonian marine calcite (−4 to −6 % VPDB [49]). Most of the calcitic samples show δ 13 C values, in range with postulated values for Devonian marine calcite with the exception for late BKC which clearly shows a negative shift from the equilibrium values ( Figure 11). This, as already mentioned for Silurian BKC, may suggest a separate input of oxidized organic carbon possibly related to interaction with hydrocarbons (e.g., [77]).
Considering that dolomite formed in equilibrium with seawater, it should display δ 18 O isotopic composition enriched of about 3 ± 1% (e.g., [82]), and assuming that Devonian seawater, with an isotopic composition of −2% SMOW [51,88], represents the source of dolomitizing fluids for the formation of RD1, its range of δ 18 O values should vary from −1 to −3% VPDB. The calculated average value (−2% VPDB) would correspond to an estimated value of 33 • C employing the fractionation equation proposed by Land [82]. However, RD1 shows a negative shift of about 1-5% from the average value (−2% VPDB) of δ 18 O expected for Devonian dolomite (Figure 10). RD2 and RD3 also exhibit a similar negative shift compared with the δ 18 O average value of Devonian marine dolomite ( Figure 10).
This negative shift of δ 18 O values can be also related to the interaction with meteoric water or formation in burial regime at relatively high temperatures (e.g., [69]). As already mentioned in the previous section, the interaction with meteoric water should be demonstrated by evidence of subaerial exposure such as karst features, such as the presence of vadose cement or other evidence associated with meteoric diagenesis. Although there is no evidence of karstification in the samples collected and analyzed in this study, a previous study [89] demonstrated the presence of dissolution processes mostly active in the shallow subsurface, usually <200 m depth in south-western Ontario. At the local scale, approximately 180 m of Devonian bedrock beneath the Bruce nuclear site are karstic with karstification and higher-permeability confined intervals also found within the Salina A1 dolostone and the Guelph Formation [89]. Hence, the possibility of meteoric water influence cannot be discounted completely.
However, the absence of a negative shift of δ 13 C values, commonly associated with the meteoric diagenesis (e.g., [69]), coupled with the marked δ 18 O isotopic negative values for matrix dolomite may favour the hypothesis of recrystallization.
The majority of the dolomite matrix, as well as the late calcite (BKC) from Devonian formations, are characterized by 87 Sr/ 86 Sr ratios falling in the range of postulated values for Devonian seawater (0.7078-0.7080 [51]) ( Figure 12). However, one sample shows more radiogenic isotopic composition than those of coeval seawater. This variation from the postulated values, even if isolated, can be related to the recrystallization of early dolomite and to water/rock interaction with the overlying Upper Devonian sandstones and shales [56]. Hence, the results 87 Sr/ 86 Sr ratios for dolomite matrix (RD1, RD2 and RD3), and late-stage calcite cement (BKC) suggest for both seawater-dominated parental fluids.

Constraints from Fluid Inclusions
As shown in Figure 15 Silurian replacive dolomite (RD2) is characterized by slightly higher homogenization temperature (89.1 • C) compared to those hosted in Devonian RD2 (83. These results coupled with the negative shift of δ 18 O values may suggest, in both age groups, the formation of RD2 was at relatively high temperatures from saline brines (e.g., [77]). Similar ranges of homogenization temperatures characterize blocky calcite cement and saddle dolomite, in the Silurian formations, and may represent evidence of similarity in terms of mineralizing fluid responsible for their formation. This assumption is also supported by evidence from REE analysis (discussed in the following section). The overall higher salinity ( Figures 15B and 16B) which characterizes RD2 and BKC from Silurian formations compared to those of the same mineral phases from Devonian formations, suggests that the diagenetic fluids responsible for their formation were saline basinal brines resulted from the dissolution of evaporites hosted in Silurian formations (e.g., [5,77]). The high homogenization temperatures of late calcite cement and saddle dolomite and the relatively high T h of fluid inclusions hosted in replacive dolomite ( Figures 14A and 15A) are significantly higher than the maximum burial temperatures suggested for the study area (ca 60-90 • C [91]) for both age groups and hence, it may suggest the involvement of hydrothermal fluids [5,13,90].
The distinct patterns of salinity and homogenization temperature (Figures 14-16) that characterize Devonian and Silurian dolomite may suggest that dolomitization/dolomite alteration and other related diagenetic processes are the result of two different diagenetic fluid systems: (1) a diagenetic fluid system that affected Silurian carbonates characterized by high temperatures and high salinity and (2) a diagenetic fluid system that affected Devonian carbonates characterized by slightly lower temperatures and salinity (e.g., [77]).

Stoichiometry
Most dolomite samples, with some exceptions for RD1, show Ca-rich, non-stoichiometric composition in both Silurian and Devonian formation. Several studies on dolomite [50,68,79,92,93] suggested that the recrystallization of fine crystalline dolomite forms coarser, more stable and more stoichiometric dolomite. However, coarser pervasive dolomites are overall less stoichiometric than fine crystalline dolomite. A possible explanation for such a trend is that these results reflect the average composition of the cores and rims of the crystals. Separate measurements for cores and rims, which were not performed for these dolomites, may have shown very substantial stoichiometric differences. The model proposed by Sibley [92] suggests that progressive recrystallization of non-stoichiometric dolomite should result in an increasingly stoichiometric dolomite characterized by Ca-rich cores and stoichiometric rims. Even though, this phenomenon has been observed and documented in many studies (e.g., [72,94]), stoichiometry has not been achieved in the Silurian and Devonian formations investigated. An experimental study by Malone et al. [95], demonstrated that dolomite subjected to temperatures of 50 • C-200 • C lead to its recrystallization with a rapid increase of stoichiometry in the first part of the experiment, but with a slowing down as reactions progressed. Despite complete recrystallization at 200 • C, dolomite stoichiometry was not achieved.
The interpretation that better explains this trend is that earlier-formed dolomite has been recrystallized by later, Ca-rich, and warmer basinal fluids, under conditions that did not allow these dolomites to achieve stoichiometry (e.g., [69]).

Trace Elements: Iron, Manganese and Strontium
The distribution of iron and manganese in solution is strictly controlled by redox conditions (Eh) [96] as well as their concentrations in precursor limestone and precipitation rate [97]. In the condition of positive Eh (oxidizing conditions), Fe 2+ and Mn 2+ oxidize passing from a divalent to a trivalent state inhibiting their incorporation in carbonates crystal lattice [59]. RD1, RD2 and SD in Silurian formations and RD1, RD2 and RD3 in Devonian formations show a progressive increase in Fe and Mn concentration (Table 5). Previous studies on dolomite [65,98] suggested that the relative enrichment of Fe and Mn in later dolomites can be attributed to dolomite recrystallization. Higher content of Fe and Mn in coarser dolomites could be related to either formation under relatively more reducing conditions [99] or increasing Fe and Mn distribution coefficients resulting from higher temperatures or slower precipitation rate [100]. Despite this increase of Fe and Mn there is no evidence of a corresponding variation in terms of Sr concentration between RD1 and younger dolomite phases in both Silurian and Devonian formations. This evidence may suggest that the different phases of dolomite observed within both age groups are perhaps linked to their parental mineralizing fluids (in this case seawater-dominated fluids). The low Sr concentrations (mostly in the range between 50 and 100 ppm for all dolomite samples in both age groups) are comparable to those of ancient dolomite precipitated from marine waters (e.g., 40-150 ppm [101]), but significantly lower than modern protodolomites (500-700 ppm) formed in evaporitic marine settings [79,102]. Decrease of Sr concentration with increasing crystal size is interpreted to be associated with dolomite recrystallization in presence of Sr-depleted meteoric waters (e.g., [50,79,98,103]. However, matrix dolomites from both age groups do not show evidence of a decrease of Sr concentrations with increasing crystal size. Despite, the lack of evidence of co-variation of Sr with either Fe and Mn, their higher concentrations in coarser dolomite (RD2 and SD in Silurian samples and RD2 and RD3 in Devonian samples) compared to RD1, can be related to their formation from Fe-and Mn-enriched fluids under more reducing conditions [7].

Rare Earth Elements (REE)
Previous studies focused on REE compositions of carbonate rocks suggest controversial scenarios in terms of preservation [104,105] or modification [54], after diagenesis, of rare earth element patterns and concentrations compared with those of their precursor seawater. Concentrations of REE in ancient carbonate rocks can be useful tools to establish the nature of the mineralizing source [106][107][108]. Chaudhuri and Cullers [109], Webb et al. [104] and Allwood et al. [105] suggested that during the post-depositional processes such as diagenesis and metamorphism REEs patterns can remain essentially unaffected, considering that a large amount of diagenetic fluids is required to record substantial changes [106]. In other words, the REE concentrations in carbonate rocks are principally influenced by the depositional environment [110,111]. However, according to recent studies (e.g., [5,54]) diagenetic processes may sensibly alter REEs patterns and concentrations, making them an ideal tool for the understanding of REEs relocation in carbonate rocks.
The distinctive Post-Archean Australian Shales (PAAS) normalized seawater REE concentrations have patterns characterized by the following features: depletion of light REE, negative Ce anomaly and slight positive La and Gd anomalies [43,[112][113][114].

Silurian
Average shale-normalized REE patterns of RD1 and RD2 show similar trends compared to those of Silurian brachiopods [54] but with significantly higher ΣREE ( Figure 17A). All dolomite and calcite samples are characterized by a slight enrichment of LREE over HREE and a more significant enrichment of MREE. The comparison between the trend of Silurian brachiopods [54] and those of RD1 and RD2 ( Figure 17A), possibly suggests an initial common seawater-dominated source of diagenetic fluids partially masked afterward by a different evolution of the diagenetic fluids [5]. Differently from Devonian patterns, Silurian dolomite trends are, in fact, characterized by a minor negative La anomaly and both cases of positive and negative Ce anomalies ( Figure 17A,B) possibly implying their formation from evolved seawater. Both types of dolomite RD1and RD2 exhibit higher average ΣREE (11.07 ± 0.93 ppm, and 9.79 3 ± 7.8 ppm, respectively) compared with the average of Silurian brachiopods (2.21 ppm) proposed by Azmy et al. [54].
Saddle dolomite (SD) and Blocky calcite cement (BKC) show similar REE shale-normalized trends ( Figure 17A) with average ΣREE (47.06 ± 41.95 ppm and 23.74 ± 23.00 ppm, respectively) and both significantly differ from the Silurian Brachiopods REE patterns. SD and BKC both show major negative La and Ce ( Figure 17B) anomalies as well as a slight negative Eu anomaly ( Figure 17A). The differences between replacive dolomite matrix RD1 and RD2 compared with Saddle dolomite and blocky calcite cement include progressively higher ΣREE and a more pronounced enrichment in LREE, and hence may suggest a different source of REEs ( Figure 17).

Devonian
Average shale-normalized REE patterns and their comparison with those of modern warm water brachiopods [54] show that all dolomite types and late calcite cement sampled from Devonian formations have maintained the original seawater REE characteristics with a negative Ce anomaly ( Figure 18A,B), and slightly positive La and Gd anomalies [43,[112][113][114]. Proceeding from older (RD1) to younger (RD3) dolomite, shale-normalized REE show similar patterns coupled with a progressive enrichment in average ΣREE (Figure 18). These characteristics may indicate a common seawater-dominated diagenetic fluid slightly modified during progressive deeper burial because of the increasing water/rock interaction [5].

Evolution of Diagenetic Fluids
The evolution of diagenetic fluids involved during dolomitization in the eastern side of the Michigan Basin can be deduced by integrating petrographic, isotopic and fluid inclusions data.
The relationships between the oxygen isotopic composition of the mineral phase, of the fluid (δ 18 O fluid ) and temperature of formation (T h from fluid inclusions studies) are shown in Figure 21. Figure 21A shows the relationship between the measured δ 18 O values of dolomite from Silurian and Devonian formations vs. temperature. It also shows the δ 18 O values of formation fluids as related to the dolomite-water oxygen fractionation equation [82]. Isotopic compositions of fluid involved in the formation of RD2 in Silurian and Devonian formations show an overlap but Devonian RD2 shows slightly more enriched δ 18 O SMOW likely related to the presence of more evolved basinal fluids (e.g., [77]) and a narrow range of temperature compared to the Silurian RD2. On the other hand, Saddle dolomite shows more enriched δ 18 O SMOW and higher temperatures than the RD2 matrix dolomite. Considering the most negative δ 18 O value from Silurian RD2 (−7.96% VPDB) and assuming a Silurian seawater value of −3.5% SMOW [83], results in a precipitation temperature of 56 • C, which is significantly lower than the average temperature obtained from fluid inclusion microthermometry (89 • C). Considering the most negative δ 18 O value from Devonian RD2 (−7.66% VPDB) and assuming a Devonian seawater value of −2% SMOW [51], results in a precipitation temperature of 33 • C, which also results significantly lower than the average temperature obtained from fluid inclusion microthermometry (83 • C). Thermal history reconstructions of the Michigan Basin have been carried out by many authors using different methods [4,6,7,13,91,115,116]. Considering the burial curve reconstructed by Zheng [7] the maximum burial temperature expected for Silurian formations would be ca. 63 • C, assuming the atmospheric temperature of 20 • C, a thickness of 1700 m for the overlaying sedimentary cover and a geothermal gradient of 25 • C /km. Recrystallization with the involvement of hydrothermal fluids (e.g., [5]) can be assumed considering the significantly higher (SD: 125 • C; RD2: 89 • C) temperature obtained from fluid inclusions microthermometric analysis. Considering the burial reconstruction proposed by Legall et al. [91] for Devonian formation using conodont and acritarch alteration index, the maximum burial temperature is estimated to be 60 • C, and hence, considerably lower than the temperatures obtained from fluid inclusions microthermometry. It follows that in both age groups, higher values of homogenization temperatures compared to those expected for maximum burial cannot be justified by the burial dolomitization model only [77] but must involve the presence of hydrothermal fluids [5,56]. Figure 21B shows the relationship between the measured δ 18 O values of late calcite cement (BKC) from Silurian and Devonian formations vs. temperature. It also shows the δ 18 O values of formation fluids as related to the calcite-water oxygen fractionation equation [117]. Isotopic compositions of fluid involved in the formation of BKC in Silurian and Devonian formations show an overlap. Devonian BKC, however, shows slightly more enriched δ 18 O SMOW values compared to the Silurian BKC. This trend reflects a slight change in fluid chemistry in the two age groups that can be attributed to more evolved basinal fluids (e.g., [77]).
from Silurian and Devonian formations vs. temperature. It also shows the δ 18 O values of formation fluids as related to the calcite-water oxygen fractionation equation [117]. Isotopic compositions of fluid involved in the formation of BKC in Silurian and Devonian formations show an overlap. Devonian BKC, however, shows slightly more enriched δ 18 OSMOW values compared to the Silurian BKC. This trend reflects a slight change in fluid chemistry in the two age groups that can be attributed to more evolved basinal fluids (e.g., [77]). The fluid flow in intracratonic basins, such as the Michigan Basin, can be influenced by topography, compaction and convection related to density gradients (e.g., [118]).
The most likely fluid flow model, considering the low topographic relief and the carbonate rocks-dominated characteristic of the Michigan Basin, is the buoyancy-driven model related to temperature and density gradient [5]. Fractures and faults in sedimentary basins characterized by alternating carbonate rocks with variable permeability can promote the migration of cross-formational fluids [119]. In the Michigan Basin, faults and fracture systems in Paleozoic successions are controlled by the reactivation of basement structures and represent preferential pathways that promote fluid migration through low permeability sedimentary rocks such as shales and evaporites [5]. Homogenization temperatures of fluid inclusions hosted in dolomite samples from both Silurian and Devonian formations are significantly higher than temperatures estimated for maximum burial but lower than temperatures of fluid inclusions hosted in dolomites from previous studies carried out in the central part of the Michigan Basin (e.g., [13,91]). This evidence suggests that dolomite formation can be related to the presence of hydrothermal fluids that originated in the central part of the Michigan Basin migrated towards the margins [5]. The potential source of heat could be related to the reactivation of the buried mid-continental rift (MCR) during late Devonian to Mississippian time [120,121], while the preferential path for the migration of diagenetic fluids from the center of the Michigan Basin can be attributed to the Cambrian sandstones unconformably overlying the Precambrian basement [5,77]. Late calcite cement in sub-horizontal fractures near the base of the Silurian sequence dated in a recent study [120] show U-Pb ages of 318 ± 10 Ma determined by LA-ICP-MS and 313 ± 1 Ma by ID-TIMS. Hence, the authors suggest that the fluid flow of hydrothermal brines was influenced by the Alleghanian orogenies [120]. According to Sutcliff et al. [122] absolute dating of Cambrian and Silurian secondary minerals shows the same age and are contemporaneous with Alleghanian magmatism and uplift which caused the migration of hydrothermal fluids westwards from high-level Alleghanian plutons through the basal Cambrian.

Conclusions
Examination of cores, petrographic studies, microthermometric and geochemical analyses of Silurian and Devonian carbonates of the eastern side of the Michigan Basin (Huron Domain) led to the following conclusions regarding dolomitization and other related diagenetic processes observed in these successions: Silurian and Devonian Sr isotopic ratios show the seawater composition of their respective age as the primary source of diagenetic fluids with minor rock-water interactions. However, Silurian samples show a more radiogenic signature that documents more intense interactions of water with feldspar-bearing rocks • REE shale-normalized patterns suggest that in both age groups the diagenetic fluids were originally of coeval seawater composition subsequently modified via water-rock interaction. The different evolution of the diagenetic fluids is more prominent in REE SN patterns from Silurian samples and it is possibly related to brines, which were modified by the dissolution of Silurian evaporites from the Salina series.