3 The Paleozoic evolution of the Appalachian orogen in New England and adjacent Atlantic Canada involved the progressive accretion to ancestral North America (Laurentia) of the outboard Gander, Avalon, and Meguma peri-Gondwanan plates (Fig. 2a), resulting in the current lithotectonic framework (Fig. 2b). Ganderia is the largest accreted plate (Fig. 2b), and studies of Ganderian volcanic units in the northern Miramichi, Elmtree, and Popelogan inliers in New Brunswick have yielded a detailed picture of subduction and associated opening and closing of back-arc basins in the central part of the plate (Table 1; van Staal et al. 2016). Previous tectonic models have focused on the nature and sequence of plate accretion events. This paper complements and helps evaluate those models, but also makes possible comparison of volcanism along the length of the Miramichi inlier.

4 Figure 3 shows relationships of the Miramichi terrane with younger cover rocks of the Central Maine/Aroostook– Matapedia (CMAM) basin to the northwest and the Fredericton trough to the southeast. The Miramichi study area shown in Figure 3 has experienced only lower greenschist facies regional metamorphism (chlorite and sub-chlorite zones) and thus represents a supracrustal level of the northern Appalachians (Ludman 2020). Delicate primary features are generally well preserved, including shard outlines in Miramichi volcanic rocks described below.

5 The Miramichi terrane in Maine is divided into two segments by the ca. 380 Ma Bottle Lake plutonic complex: the Danforth segment between the complex and the New Brunswick border, and the smaller Greenfield segment southwest of the complex (Fig. 3). The Pokiok igneous complex at the Maine–New Brunswick border further separates the Maine part of the inlier from the southwestern New Brunswick components.

6 Regional-scale faults separate the Miramichi terrane from adjacent cover rocks. The southeastern contact with the Fredericton trough is the Codyville fault in the Danforth segment, northwesternmost branch of the transcurrent Norumbega fault system, and the Southeast Boundary Fault, its possible continuation, in the Greenfield segment. The evolution of the western contact was complex, with the Northwest Boundary, Stetson Mountain, and North Bancroft faults locally separating Miramichi from CMAM strata. Evolution of these boundaries and their relationships with faults in New Brunswick is beyond the scope of this paper; for details see Ludman (in press).

7 Despite its tectonic significance, the Miramichi terrane in Maine received little attention from previous investigators. The first modern report was a United States Geological Survey regional reconnaissance study of central eastern Maine that included the Danforth segment but did not extend as far southwest as the Greenfield area (Larrabee et al. 1965). That study briefly described volcanic rocks to which a Middle to Late Ordovician age was assigned based on graptolites in a black shale horizon just north of Danforth. It also suggested similarities with the Tetagouche volcanic section in northern New Brunswick and with rocks of the Weeksboro– Lunksoos Lake belt in northern Maine. Detailed mapping between 1980 and 2020 established and refined Danforth segment stratigraphy and interpreted a complex, multideformation tectonic history described below (Ludman 1985, 1991, 2003, 2020; Ludman and Berry 2003; Ludman and Hopeck 2020).

8 Sayres (1986) suggested an internal stratigraphy for Miramichi volcanic rocks in the Danforth 15′ quadrangle (see below) and reported chemical analyses revealing a calcalkaline suite of andesite, rhyodacite, and rhyolite that she interpreted as a volcanic arc assemblage (Sayres 1986). A study of the Meductic arc in southwestern New Brunswick included a few samples from the Danforth area, and supported Sayres’ tectonic interpretation (Winchester et al. 1992).

9 The Greenfield segment was first studied by Olson (1972) in the southwestern part of what is now the Greenfield 7½′ quadrangle. His Masters thesis described a thick volcanic section composed mostly of andesitic and rhyolitic pyroclastic rocks but provided no chemical data. He tentatively correlated the volcanic rocks with those in the Danforth area and with the Weeksboro–Lunksoos Lake section based on descriptions in Larrabee et al. (1965) and Neuman (1962), respectively. Greenfield segment stratigraphy has been revised and volcanic geochemistry reported in recent studies of the Greenfield 7½′ quadrangle (Ludman 2020, in press).

10 A well-defined stratigraphy in the Danforth segment is truncated by the Bottle Lake pluton, but its basal Baskahegan Lake and Bowers Mountain formations reappear at the northern margin of the Greenfield segment (Fig. 4). The Bowers Mountain is absent in the southern part of the Greenfield segment, where a unique, thin-bedded Lazy Ledges Road member of the Baskahegan Lake Formation passes upward into volcanic rocks of the Olamon Stream Formation.

11 Thick volcanic suites cap the Miramichi section in both segments and, although similar, are different enough to require separate names – the Stetson Mountain Formation in the Danforth segment, and the Olamon Stream Formation in the Greenfield area. The Olamon Stream interpretation presented here is a revision of a recent report (Ludman 2020).

12 Sayres (1986) divided the Stetson Mountain Formation into a thin basal member of cryptocrystalline to fine grained felsic and intermediate ashfall tuff and a much thicker upper member of coarser felsic and intermediate ashflow tuff and agglomerate, separated by a thin, medial member of chalky weathering, black sooty iron- and manganese-rich volcanogenic (?) mudstone. A similar sequence appears to occur in the Olamon Stream Formation although outcrop control of the lower member is very poor. The lower member comprises mostly cryptocrystalline tuff, separated by an anoxic mudstone/siltstone (Greenfield member) from the upper member which appears to contain more abundant coarser pyroclastic rocks. The most significant difference between the two segments is the presence of relatively coarse grained mafic volcanic (and subvolcanic?) rocks in the Greenfield segment. Unfortunately, the stratigraphic position of the mafic rocks within the Olamon Stream Formation is unknown.

13 The Danforth and Greenfield suites contain predominantly (>95%) pyroclastic rocks, a variety of intermingled intermediate (andesite, basaltic andesite) and felsic (dacite, rhyodacite, rhyolite) crystal, lithic, and crystal-lithic ashfall and ashflow tuffs, breccias, coarse agglomerates and subordinate lavas with rare volcaniclastic sedimentary horizons. Mafic rocks are largely confined to a small area at the southwest terminus of the Miramichi terrane, with sparse basaltic tuffs elsewhere in the Olamon Stream Formation. The most mafic rock in the Danforth segment is a low-silica basaltic andesite that occurs only as xenoliths in the Bottle Lake pluton and in a small fault-bounded sliver between that pluton and the Baskahegan Lake Formation.

26 Seventeen samples of Olamon Stream volcanic rocks were collected in 2019 and sent to Activation Laboratories Ltd. in Ancaster, Ontario, for preparation and analysis of major elements by induction coupled plasma optical emission spectroscopy, and trace elements by induction coupled plasma mass spectrometry. Fifteen samples of least altered volcanic rocks from the Danforth segment had been collected by Sayres (1986), prepared at Queens College, and analyzed at X-ray Assay Laboratories Ltd (Don Mills, Ontario) by X-ray fluorescence. Five additional samples from the Danforth segment were also processed by Activation Labs in 2014.

27 Thirty-eight samples were selected for major element geochemistry from the volcanic rocks in the Greenfield and Danforth segments of which 18 contain trace element compositions and 22 have rare earth element compositions (Appendix A). For samples collected in this study the Universal Transverse Mercator (UTM) locations are recorded in Appendix B. Locations for previous samples collected from the Danforth segment are recorded in Sayres (1986). Volcanic rock types range from basalt to basaltic andesite, andesite, dacite, and rhyolite (Fig. 12), mirroring the textural variety shown in Figures 9, 10, and 11. Overall, rocks in the two segments exhibit similar compositions and proportions although, as mentioned earlier, mafic rocks present in the Olamon Stream Formation are absent from the Stetson Mountain Formation. At this time, several explanations are plausible for this difference: different eruptive histories at different volcanic centers; different erosional levels in the two segments (although the bases of both can be observed, the top of neither is exposed); and structural complexities.

28 Sayres (1986) interpreted an internal stratigraphic sequence based on texture and eruptive style in the Danforth segment in an area with excellent outcrop control. However, no systematic compositional trend is apparent in the Stetson Mountain Formation, as andesite, dacite, and rhyolite are found in similar abundance in the lower and upper parts of the formation, commonly in the same outcrop. Similar combinations of intermediate and felsic rocks at individual outcrops are also common in the Olamon Stream Formation (e.g., Appendix A, samples 18B37a, b).

29 Nine representative Miramichi volcanic rocks were collected for U–Pb dating in the summer of 2019, three each from the mafic member and main body of the Olamon Stream Formation, and three from the Stetson Mountain Formation. These samples were sent to Overburden Drilling Management, Ltd., Ottawa, Canada, for electric pulse disaggregation and initial zircon separation. Further separation and hand picking were carried out at the University of New Brunswick geochronology laboratory in Fredericton, New Brunswick, Canada.

30 U–Pb analyses were conducted at the University of New Brunswick using an ASI M-50 193 nm excimer laser ablation system (Complex Pro 110) with a 20 ns pulse width. The ablation system is equipped with a Laurin Technic Pty S–155 two–volume ablation cell coupled, using 4mm OD Nylon tubing, to an Agilent 7700× quadrupole ICP-MS equipped with dual external rotary pumps. Ablated material was transported to the ICP-MS using a mixed He (300 mL/ min) and Ar (930 mL/min) carrier gas. Sensitivity was enhanced using the second external rotary pump and adding N₂ (2.0 mL/min) downstream of the cell. Due to isobaric interference of ²⁰⁴Pb with ²⁰⁴Hg impurities within the carrier gas, direct measurement of the ²⁰⁴Pb signal was permitted by using in-line high-capacity Vici Metronics Hg traps on all gas lines ensuring that ²⁰⁴Hg remains <150 cps under maximum sensitivity conditions. Correction of the ²⁰⁴Hg interference on ²⁰⁴Pb was then performed via peak–stripping using the measured ²⁰²Hg on the gas background and assuming a canonical value for ²⁰²Hg/²⁰⁴Hg.

31 Analyses of all unknowns were bracketed with at least 16 spots on primary reference material to monitor and correct for instrument drift and time-dependent down-hole fractionation. For zircon, FC-1 (ca. 1099 Ma; Paces and Miller 1993) was used as primary external standard and accuracy was checked using Plesovice (ca. 337 Ma; Sláma et al. 2008 ) or GSC z1242 (ca. 2679 Ma; B. Davis personal communication 2021) A 17 µm crater size, 3 J/cm² laser energy, and a repetition rate of 4 Hz were used for all unknowns and standards. ⁹⁰Zr was used as a guide mass for ablations and internal standardization using an estimated value of 47 wt% Zr. NIST 610 was used for instrument tuning and as a concentration standard. Drift was modeled using Iolite3.7^TM automatic fit function whereas down-hole fractionation was modeled using the automatic exponential fit option. The net ²⁰⁴Pb (cps) signal was used for common Pb corrections (if required) for zircon and incorporated into the VizualAge U–Pb data reduction scheme (DRS) (Petrus and Kamber 2012) operating under Iolite3.7^TM. The routine uses an age estimate based on the measured ²⁰⁶Pb/²³⁸U and common Pb ratios based on the evolution curves of Kramers and Tolstikhin (1997). Common Pb corrections were applied only to zircon analyses with net ²⁰⁴Pb >50 cps and 1σ internal errors <30%. All other data remained uncorrected.

32 Three volcanic rocks from the Olamon Stream Formation upper member yielded enough zircon grains for valid dating (Table 2). They include only two of the nine collected in 2019, dated by LA-ICPMS, and one dated earlier by U–Pb SHRIMP (Ludman et al. 2018). Concordia diagrams and images of zircons from the two new ages are shown in Figures 13 a and b. The UTM locations and the complete analytical data for the new dates are given in Appendix B and C. Unfortunately, none of the Danforth segment samples could be dated and the age of the Olamon Stream basalts is still uncertain.

33 The three ages are essentially identical, indicating eruption of the upper member of the Olamon Stream Formation at about 469 Ma, the boundary between the Lower and Middle Ordovician series (Floian–Dapingian stages). Unfortunately, they offer no insight into the timing of initial volcanism or the internal stratigraphy of the formation.

34 Attempts to date Olamon Stream basalts were inconclusive. Only one sample(20A8) had enough zircon grains to be dated, and most of these grains yielded a range of inheritance ages (Fig. 13c). Two zircon analyses close to Concordia suggest an age similar to those of the dated samples but are not conclusive.

36 Miramichi volcanic rocks in Maine are a classic calcalkaline basalt-andesite-rhyolite suite (Fig. 14a), associated with a subduction-related volcanic arc (Fig. 14b), consistent with previous interpretations of Miramichi volcanism (e.g., Sayres 1986; Winchester et al. 1992; van Staal et al. 2016). Niobium, lanthanum, and ytterbium data further suggest eruption in an ensialic continental arc rather than an oceanic setting (Fig. 14c). Lava flows, ashfall and ashflow tuffs, and volcanic agglomerates indicate recurrent Plinian and phreatomagmatic eruptions consistent with this environment.

37 Stratigraphic, lithologic, chemical, and geochronological data presented above permit correlation of the Olamon Stream and Stetson Mountain formations with volcanic rocks of the Miramichi inlier in New Brunswick and the Weeksboro–Lunksoos Lake and Munsungun Cambrian– Ordovician inliers in northern Maine. Direct connections are not possible with northern Maine inliers because of the extensive cover rocks, nor with the New Brunswick Miramichi volcanic rocks because the Pokiok plutonic suite at the New Brunswick–Maine border separates the Danforth segment from potential correlatives in west-central New Brunswick.

38 Structural complexities also obscure Maine–New Brunswick correlation. The Miramichi inlier is separated everywhere from adjacent Late Ordovician to Late Silurian strata by regional-scale boundary faults and is cut into discrete blocks by internal faults that had complex thrust, high-angle dip-slip, and strike-slip episodes (Fig. 15; Ludman in press). The Bottle Lake and Pokiok complexes also interrupt continuity of these faults, adding another element of uncertainty to correlations. Indeed, depending on how the faults are connected through the Pokiok plutonic suite, otherwise strong candidates for correlation may lie in different fault blocks.

39 Thus, although areas of Early to Middle Ordovician Miramichi volcanic rocks in Maine and New Brunswick are parallel to the fault-controlled trend of the terrane, their component rocks may not actually be on strike with one another. Figure 15 shows one of several possible connections of Miramichi faults in Maine and New Brunswick. This model was chosen so that the Greenfield and Danforth segments are within the same fault block, but as a result, the Meductic Group in the Eel River are lies on another.

64 Models for the timing and nature of events at the leading edge of Ganderia that first collided with Laurentia have been based largely on information from New Brunswick: the Miramichi and Elmtree inliers; the much smaller Popelogan inlier and even smaller Oxford Brook exposure to the west; and the Straughan Brook drill hole through the Carboniferous Maritimes Basin to the east (Fig. 1). Maine Miramichi volcanic data presented here, combined with new information from the extensive Weeksboro–Lunksoos Lake and Munsungun belts closer to that leading edge (Fig. 1), add significantly to our understanding of Ganderian accretion to Laurentia.

65 As described above, our geochemical data support the model of continental arc volcanism attributed to the Popelogan arc, and the Poplar Mountain volcanic suite is consistent with an intra-arc or back-arc rifting setting (Tetagouche Group), and subsequent closing of the back-arc basin (van Staal et al. 2016). That model proposed initial rollback of an eastward-dipping subducted slab beneath the Popelogan arc, accompanied by northwestward migration of arc volcanism, first to the Weeksboro–Lunksoos Lake inlier and eventually as far as the Munsungun inlier (Figs. 21a, b).

66 This model, however, is not consistent with simultaneous Floian–Dapingian volcanism in the three Maine inliers described above (Fig. 22), including at ca. 471–468 Ma in the Miramichi, Weeksboro–Lunksoos Lake and Munsungun inliers, and at ca. 480–484 Ma in the Meductic Group and possibly ca. 485–481 Ma in the Weeksboro–Lunksoos Lake inlier (Table 3). The current distance between the Miramichi and Munsungun inliers is about 150 km and although it is difficult to estimate the original span because of Ordovician extension and compression and Salinic and Acadian telescoping, it was almost certainly significantly greater than today.

67 Even the current 150 km span is much wider than can be explained by a single shallow eastward-dipping subduction zone. We therefore propose that simultaneous Ordovician volcanism in the three inliers indicates eruptions in two or perhaps as many as three subduction zones (Fig. 23), analogous to the Molucca–Celebes Sea collision zones (Hall and Nichols 1990). The largely bimodal nature of Weeksboro– Lunksoos volcanism suggests a third possibility: that it and the Munsungun rocks may represent an arc/back-arc system comparable to that of the Meductic/Tetagouche rocks.

68 The multiple arc model would be more convincing if there were associated melanges like that accompanying Penobscot arc closure in the Munsungun inlier, or subduction complexes like that in the Bathurst area. The absence of these features, however, does not preclude a subduction-related arc. For many years, their absence from the Eastport–Mascarene volcanic belt in eastern Maine and southeastern New Brunswick prevented widespread acceptance of a subduction model, but Llamas and Hepburn (2013) argued convincingly for eruption over a west-dipping subduction zone.

69 Neuman (1967) was the first to recognize an unconformity in the Weeksboro–Lunksoos Lake inlier separating early Ordovician volcanic rocks (Shin Brook Formation) from the underlying Cambrian–Ordovician Grand Pitch Formation. This unconformity is now recognized in a broad area from the Munsungun inlier in northeastern Maine (Wang 2018), to northern (Bathurst area) and southeastern (Annidale area) New Brunswick and is attributed to a Penobscot orogeny related to early Ordovician terrane accretion on the trailing edge of Ganderia (Fyffe et al. 2011; van Staal et al. 2016).

70 This early Ordovician unconformity has not been recognized, however, in the southwestern part of the Miramichi inlier in Maine and New Brunswick. Instead, there is a continuous and gradational transition upward from the Woodstock Group to the Meductic Group as described above (Fyffe 2001; McClenaghan et al. 2016). Contact relationships are less clear in Miramichi exposures in Maine, but the basal Baskahegan Lake and Bowers Mountain formations appear to have experienced the same deformation history as the overlying Olamon Stream and Stetson Mountain formations.

71 Its absence does not refute the existence of a Penobscot orogeny, nor its tectonic interpretation, but does suggest that its geographic impact was more complex than previously interpreted, and that additional work is necessary to explain its absence.

72 This paper is a report of progress made during nearly half a century of mapping in the Miramichi Cambrian–Ordovician inlier in Maine (Ludman 1978, 1991, 2003, 2020, in press) and recent studies in the Munsungun inlier (Wang 2018, 2019, 2021). Much of that work is ongoing, and future improvements in understanding their stratigraphy and relationships with the cover rocks will undoubtedly require further revisions in regional tectonic models. Increased interest in volcanic-hosted ore deposits at Pickett and Bald mountains in the Weeksboro–Lunksoos Lake and Munsungun inliers has generated useful geochemical and geochronologic data and further development of these prospects will continue to provide valuable new information.