4 The Central Maine basin underlies the western portion of the study area and represents a regionally extensive, Late Ordovician to Early Devonian assemblage of marine sedimentary rocks (Osberg 1988). The basin, as presently preserved in central Maine, is up to 100 km wide and interpreted to have originally formed between the Taconic orogeny-modified margin of eastern North America (Laurentia) and a series of outboard pre-Silurian Gondwana-derived terranes (Robinson et al. 1998; van Staal et al. 2009). In our study area (Fig. 2), only the eastern structural margin of the Central Maine basin is exposed and it contains deformed and metamorphosed turbidites of the Late Ordovician (?) to Silurian Vassalboro Group (Marvinney et al. 2010). Importantly, fossils from the Waterville Formation within the Vassalboro Group contain Llandoverian to Wenlockian fossils (Pankiwskyj et al. 1976; Orr and Pickerill 1995). Lithologically similar but unfossiliferous rocks of the Hutchins Corner and Mayflower Hill formations conformably underlie and overlie the Waterville Formation, respectively (Marvinney et al. 2010). Where the stratigraphic position of rocks relative to the Waterville Formation cannot be ascertained, they are simply assigned to the Vassalboro Group. One sample (SC-9) was collected and analyzed from the easternmost part of the belt preserved in the study area. Bradley and O’Sullivan (2017) recently completed a detailed detrital zircon study of strata from the northwestern margin of the Central Maine basin, and Ludman et al. (2018) present detrital zircon data from the easterly portions of the basin. The results of these studies will be compared with our findings later in the paper.

5 The Liberty-Orrington belt is exposed immediately to the southeast of the Central Maine basin and is a relatively narrow (< 25 km), approximately 175 km-long belt that contains metamorphosed Middle to Upper Ordovician volcanic and volcanogenic sedimentary rocks of the Falmouth-Brunswick Group and Casco Bay Group (Hussey et al. 2010). In south-central Maine, these rocks are interpreted to represent an assemblage of volcanic arc to backarc rocks of Gondwanan affinity (i.e., Ganderia) that were accreted to the Laurentian margin during the Silurian (West et al. 2003, 2004). One sample (SC-8) analyzed in this study was collected from the Cape Elizabeth Formation of the Casco Bay Group. Metasedimentary rocks of the Cape Elizabeth Formation are conformably overlain by metavolcanic rocks of the Spring Point Formation – the protoliths of which were erupted at 469 ± 3 Ma (based on a U–Pb zircon crystallization age from Tucker et al. 2001). Detrital zircon ages from the Wilson Cove Member of the underlying Cushing Formation are reported in West et al. (2008).

6 The Fredericton trough in south-central Maine consists of metamorphosed metasedimentary rocks of the Bucksport Formation, Appleton Ridge Formation, and Ghent Phyllite. These rocks have been tentatively correlated with Silurian rocks of the Kingsclear Group in southwestern New Brunswick (McKerrow and Ziegler 1971; Fyffe et al. 2011). In this study, detrital zircon from one sample of the Appleton Ridge Formation (SC-6) and one sample of the Ghent Phyllite (SC-7) were analyzed. Detrital zircon ages have recently been published from various formations in the Kingsclear Group in New Brunswick (Dokken et al. 2018) and adjacent eastern Maine (Ludman et al. 2018); these results will be discussed later in the paper.

7 The Jam Brook Complex, located just to the east of the Sennebec Pond fault, contains a wide variety of relatively thin (<100 m) fault-bounded metasedimentary units contained within a belt up to 2 km wide and 15 km long. These rocks were originally interpreted by Bickel (1976) to represent a single stratigraphic unit (the Jam Brook Formation), but they are currently interpreted to represent different stratigraphic units juxtaposed by a complex series of strike-parallel faults (Berry et al. 2016). Rock types within what is now viewed as a structural complex associated with the Sennebec Pond fault include quartzite, metaconglomerate, marble, calc-silicate gneiss, mica schist, and felsic gneiss. Depositional age control on these individual rock types is lacking, but lithologic similarities suggest correlations with stratigraphic units to the east (e.g., parts of the Rockport Group and Cookson Group) suggesting an age range of Precambrian to Ordovician for the various rock types within the complex (Berry et al. 2016). Detrital zircon from two quartzite samples (SC-4 and SC-5) within the Jam Brook Complex were analyzed in this study. No other age control (fossils or geochronology) is available from rocks of the Jam Brook Complex.

8 The St. Croix belt extends over 200 km from the present study area along the western side of Penobscot Bay through eastern Maine (Ludman 1987) and into southern New Brunswick (Fyffe and Riva 1990; Fyffe et al. 2011). The belt contains the rocks of the Cookson Group, which are dominated by Middle Cambrian to Late Ordovician metasedimentary and minor metavolcanic rocks. In the study area, the Cookson Group contains rocks of the Penobscot and Megunticook formations and these units have been tentatively correlated with the Calais and Crocker Hill formations, respectively, in eastern Maine and coastal New Brunswick. U–Pb dating of igneous zircon crystallization in metavolcanic rocks from the upper part of the Penobscot Formation in the study area reveal latest Cambrian to earliest Ordovician eruptive ages (Tucker et al. 2001; Berry et al. 2016), consistent with fossil evidence in southern New Brunswick (Fyffe and Riva 1990). In addition, immobile trace element geochemistry in mafic metavolcanic rocks of the Penobscot Formation is consistent with formation in an island arc system generated through ocean-ocean subduction (Burke 2016). Unfortunately, a sample collected from the Megunticook Formation (SC-3) did not yield a sufficient number of zircon grains for analysis.

9 The Clarry Hill Formation consists of migmatitic schist contained within a thrust fault bounded klippe that is structurally above the Kingsclear Group of the Fredericton trough (Tucker et al. 2001; Berry et al. 2016); these rocks have been correlated with the Megunticook Formation of the St. Croix Group (previous references). Unfortunately, a sample collected from the Clarry Hill Formation (SC-2) did not yield a sufficient number of zircon grains for analysis. No other age control (i.e., fossils or geochronology) is available from rocks of the Clarry Hill Formation and thus proposed correlations with the Megunticook Formation remain untested.

10 The Benner Hill Sequence (Osberg and Guidotti 1974) is a spatially restricted, fault-bounded assemblage of Ordovician metasedimentary rocks that have unclear relationships with surrounding terranes (Berry et al. 2000, 2016). The sequence contains the Benner Hill, Hart Neck, and Mosquito Harbor formations and, importantly, the Hart Neck Formation contains deformed brachiopods which indicate a Caradocian depositional age (Boucot et al. 1972). One sample analyzed in this study was collected from the fossiliferous Hart Neck Formation (SC-1).

17 Sample SC-9 is a calcareous granofels (Fig. 3c) from well-bedded biotite and calc-silicate granofels of the Vassalboro Group exposed near the eastern structural margin of the Central Maine basin. Mineral separation yielded abundant detrital zircon displaying a variety of sizes, external morphologies, and internal textures (Fig. 4d). U–Pb isotopic ratios were obtained for 168 detrital zircon grains from the sample and of the 167 successful analyses, 161 plotted within 15% of concordia. The distribution of calculated zircon ages (Fig. 5a) shows prominent peaks at ~ 460 Ma, 1060 Ma, and 1600 Ma with a minor peak at ~2670 Ma. The three youngest detrital zircon dates from the sample are between 420–425 Ma.

18 Sample SC-8 is a mica- and feldspar-bearing quartzite from interlayered quartzite and sillimanite-bearing schist of the Cape Elizabeth Formation of the Casco Bay Group (Fig. 3a). Mineral separation yielded an abundance of detrital zircon displaying a variety of sizes (up to 100 µm across), external morphologies, and internal textures (Fig. 4c). U–Pb isotopic ratios were obtained for 176 detrital zircon grains from the sample and of the 175 successful analyses, 164 plotted within 15% of concordia. The distribution of calculated zircon ages (Fig. 5b) shows a very prominent peak at ~600 Ma, with broader, smaller peaks centered at ~1210 and 1920 Ma. The three youngest detrital zircon dates from the sample are 469 and 481(2) Ma.

19 Sample SC-7 is a fine-grained metasiltstone (Fig. 3d) collected from interlayered phyllite and metasiltstone of the Ghent Phyllite (garnet zone) near the eastern margin of the Fredericton trough. Mineral separation provided a large number of mostly clear, internally zoned zircons in the 50–100 µm size range. U–Pb isotopic ratios were obtained for 200 detrital zircon grains from the sample and of the 195 successful analyses, 157 plotted within 15% of concordia. The distribution of calculated zircon ages (Fig. 5c) shows a very prominent peak at ~540 Ma with a noticeable shoulder at ~600 Ma. Numerous zircons have older dates in the 900– 2200 Ma range with the larger of these small broad peaks being at ~1520 and 2090 Ma. The three youngest detrital zircon dates are 435, 468, and 480 Ma.

20 Sample SC-6 is from a cordierite-bearing, feldspathic metasandstone from interlayered feldspathic metasandstone and staurolite-bearing schist of the Appleton Ridge Formation (Fig. 3b). Mineral separation produced a large number of fragmented grains that display complex internal zoning. U–Pb isotopic ratios were obtained for 96 detrital zircon grains from the sample and of the 87 successful analyses, only 62 plotted within 15% of concordia. The distribution of calculated zircon dates (Fig. 5d) shows a very prominent peak at ~550 Ma with older, smaller broad peaks being at ~1480 and 2060 Ma. The three youngest detrital zircon dates are 422, 494, and 499 Ma.

21 Sample SC-5 is a light grey to white quartzite (Fig. 3e) displaying flattened quartz clasts up to 3 cm in length. Zircons separated from this sample showed more size and shape variability than other samples in the study, with a large proportion of the grains being <20 µm fragments which were too small to analyze. U–Pb isotopic ratios were obtained for 114 detrital zircon grains from the sample and of the 106 successful analyses, 104 plotted within 15% of concordia. The distribution of calculated zircon dates (Fig. 5f) shows two prominent peaks at ~1250 and 1550 Ma. Only two detrital zircon dates from this sample are less than 1000 Ma (960 and 969 Ma).

22 Sample SC-4 is a light grey orthoquartzite (Fig. 3f). Mineral separation yielded unusually euhedral zircon as compared with other samples in the study and xenocrystic cores are uncommon in the zircons imaged (Fig. 4b). U–Pb isotopic ratios were obtained for 120 detrital zircon grains (all successful), with 116 of these analyses plotting within 15% of concordia. The distribution of calculated zircon dates (Fig. 5g) shows a very prominent narrow peak at ~470 Ma with only two dates (952 and 1216 Ma) outside the 439–529 Ma age range associated with this peak.

23 Sample SC-1 is a rusty-weathering quartzite from the top of the Hart Neck Formation. Mineral separation produced the cleanest zircon separate of the study, which consists of grains with uniform sizes and morphologies (Fig. 4a). U–Pb isotopic ratios were obtained for 168 detrital zircon grains (all successful) with 161 of these analyses plotting within 15% of concordia. The distribution of calculated zircon dates (Fig. 5e) shows a very prominent narrow peak at ~540 Ma with a noticeable shoulder at ~600 Ma. Much smaller peaks are found at ~1470 and 1960 Ma. The three youngest dates calculated are 302, 410, and 466 Ma, though the vast majority of the dates are between 500 and 600 Ma.

24 With the exception of the Hart Neck Formation of the Benner Hill Sequence, none of the metasedimentary rocks exposed in the study area contain fossils that can be used to constrain their original depositional ages. Although U–Pb crystallization ages from conformable metavolcanic units can provide absolute age control on deposition (discussed for relevant samples below), most of the age assignments for stratified rocks in the study area are based on along-strike correlations to the northeast where the rocks have not been extensively recrystallized by the combined effects of ductile deformation and amphibolite-facies metamorphism. Berry and Osberg (1989) provided a comprehensive overview of many of the individual units in the study area and discuss likely correlatives, but depositional age assignments for many of the stratified rocks exposed in the area of Figure 2 have few constraints.

25 Detrital zircon geochronology can provide an independent means for constraining the depositional ages of sedimentary/metasedimentary rocks. These constraints are based on the “law of detrital zircons” which states that a sedimentary unit cannot be any older than the youngest zircon grain it contains (Gehrels 2014). In other words, the youngest detrital zircon age obtained from a sample provides a maximum depositional age. However, the single youngest detrital zircon age acquired from a sample should be interpreted with caution as post-crystallization Pb-loss or post-depositional zircon growth during metamorphism can lead to anomalously young ages in detrital zircon populations, resulting in underestimates of maximum depositional age (Gehrels 2014).

26 In order to determine best practices for the interpretation of maximum depositional ages from detrital zircon data, Dickinson and Gehrels (2009) compared several different methods for determining the “youngest grain” against known, biostratigraphically constrained depositional ages of Mesozoic sandstone from the Colorado Plateau. They found the greatest precision in a method that involved taking the weighted mean age (incorporating analytical and systematic errors) of the youngest cluster of three or more grains that overlapped in age within errors of 2σ. In this study, to provide an independent estimate of the maximum depositional age of a sampled unit, we have adopted the Dickinson and Gehrels (2009) approach and use a weighted mean (weighted by the inverse square of 2σ) of the three youngest grain ages overlapping in 2σ that are not considered outliers given known age constraints. The maximum depositional ages determined using this method are presented in Table 2 and discussion of additional constraints for individual units is provided below.

32 One of the more powerful interpretive tools made possible by detrital zircon geochronology is the determination of sediment provenance (e.g., Gehrels et al. 2011). By linking detrital zircon populations found in samples of sedimentary rock to known sources of igneous zircon, these interpretations can indicate the location of initial sedimentary deposition relative to sources of detrital zircon. The basis of detrital zircon provenance interpretation is the fact that for a given depositional setting, detrital zircon could be sourced from any number of different surrounding landmasses (e.g., continents and island arcs) with unique tectonic histories which introduced zircon into the sediment cycle at varying rates over time (Gehrels 2014). By matching the distribution of detrital zircon ages from a sample (i.e., peaks at particular ages) to known periods of magmatism in potential sources, a link is drawn between the crystallization of zircon in a particular region and its deposition as detrital zircon in an adjacent sedimentary basin (Thomas 2011).

33 Lithostratigraphic belts preserved in the south-central Maine portion of the northern Appalachians (Fig. 2) initially formed in the Iapetus and Rheic Oceans before being accreted to the Laurentian margin during Silurian-Devonian orogenesis (e.g., van Staal et al. 2012; van Staal and Barr 2012; Domeier 2016). Sedimentary basins in these oceans were surrounded by a number of potential sources of detrital zircon including the large continental landmasses of Laurentia and Gondwana, smaller Gondwana-derived continental landmasses (e.g., Ganderia, Avalonia, Meguma), and numerous evolving early Paleozoic volcanic arc terranes (e.g., Penobscot and Popelogan–Victoria arc terranes). Waldron et al. (2018) recently summarized many of the first-order differences between detrital zircon populations derived from these different sources. These major differences include: (1) Laurentian sources are characterized by high concentrations of detrital zircons between 900 and 1100 Ma, with a broad range of much less abundant Mesoproterozoic and Paleoproterozoic zircons (i.e., subdued peaks in an age distribution); (2) Gondwana-derived sources (including Ganderian basement), although variable depending on the specifics of origin (e.g., Ganderia versus Avalonia), have large concentrations of zircon in the 540 to 650 Ma age range, generally lack 900 to 1100 Ma zircons, and many samples have zircons in the 1900 to 2200 Ma age range; and (3) intra-oceanic volcanic arc terranes contain zircon populations corresponding to the time of volcanism in the arcs (e.g., 515–485 Ma for Penobscottian arcs and 475–455 Ma for Popelogan-Victoria arc terranes).

41 An additional interpretive method applied to detrital zircon geochronology is the determination of the tectonic setting of sediment deposition. Similar to the interpretation of sediment provenance, the technique works by exploiting known signatures in zircon age distributions. However, rather than reflecting particular events in the geologic record, the signatures used in this method result from different sets of conditions unique to each tectonic regime which affect how detrital zircon is introduced into the sediment cycle. These differences are primarily controlled by two factors: the volume and preservation potential of zircon-producing magma generated in the system, and the degree to which that magmatic material is available for erosion of detrital zircon (Cawood et al. 2012). For example, a basin developed on a convergent plate boundary (e.g., a forearc or backarc basin) is likely to receive zircon that crystallized relatively recently during arc volcanism, resulting in a small difference between the age of deposition and the crystallization age of detrital zircons. In contrast, a basin in a collisional regime (e.g., a foreland basin) is more likely to receive detrital zircon eroded from relatively old intrusive igneous bodies that only recently reached the surface, resulting in a pronounced difference between the timing of zircon crystallization and deposition.

42 These differences in crystallization ages and depositional ages can be observed in probability density distributions. For instance, the narrow, isolated peak in the age distribution from SC-4 (Fig. 5f) represents a small difference between these ages and is suggestive of sedimentary deposition in a convergent regime proximal to active zircon-producing felsic to intermediate magmatism. However, as demonstrated by Cawood et al. (2012), this metric is best visualized by a cumulative proportion curve (CPC) that shows which percentages of the analyzed zircon population correspond to particular differences between crystallization and sedimentation age. These authors compiled detrital zircon data from a number of different types of basins within three endmember tectonic settings and plotted the corresponding CPCs to identify signatures that could link other zircon age distributions to potential tectonic environments (Fig. 6a). There is overlap in the generalized regions of the plot occupied by the CPCs of each of these endmembers, but gaps between them provide two percentage thresholds at which tectonic designations can be assigned. If the difference between crystallization age and depositional age (CA−DA) is >150 Ma in 5% of the sample, this indicates an extensional setting (labeled “C” in Fig. 6a). If CA−DA is <100 Ma in 30% of the sample, a convergent regime is assigned (labeled “A” in Fig. 6a). If 30% of the sample has a CA−DA >100 Ma, it is given a collisional setting designation (labeled “B” in Fig. 6a).

43 Figure 6b presents a compilation of cumulative proportion curves associated with the detrital zircon age distributions of each sample analyzed in this study. In comparing these curves to the signature thresholds of endmember tectonic settings provided by Cawood et al. (2012) in Figure 6a, the following assignments can be made: (1) convergent settings are clearly exemplified by samples SC-1 and SC-4 from the Hart Neck Formation and Jam Brook Complex, respectively; (2) sample SC-5 of the Jam Brook Complex shows an extensional setting; and (3) collisional settings are displayed by samples from the Ghent Phyllite of the Fredericton trough (SC-7) and Cape Elizabeth Formation of the Casco Bay Group (SC-8).

44 Tectonic settings for the remaining two samples are less clear as there is a difference between the settings indicated by their ages at the 5% and 30% thresholds and the settings indicated by the overall shape of their respective CPCs. The Appleton Ridge Formation sample from the Fredericton trough (SC-6) has CA−DA <100 Ma in 30% of zircon grains suggesting a convergent setting, but the extent of its CPC fits better with a collisional setting. Similarly, the CPC for the Vassalboro Group sample from the Central Maine basin (SC-9) fails to meet the CA−DA >150 Ma threshold in 5% of grains but otherwise appears most similar to an extensional setting. This general designation is consistent with previous tectonic interpretations of the Central Maine basin (Bradley et al. 2000; Bradley and O’Sullivan 2017), although details remain unresolved (e.g., whether this extensive basin was floored by oceanic lithosphere). The ambiguous nature of these two CPCs is likely due to a combination of non-ideal sample characteristics (i.e., low n-value in SC-6 and dual sediment sources in SC-9).

45 The previously presented interpretations of depositional age, sediment provenance, and tectonic setting are summarized in Table 3 and provide key insights into the origins of the different lithotectonic belts currently juxtaposed in south-central Maine. These insights demonstrate that the bedrock geology preserved in this region resulted from the accretion of multiple terranes containing sedimentary rocks that formed in a variety of basins proximal to several distinct sediment sources over the course of hundreds of millions of years. Pre-Silurian rocks (Casco Bay Group, Benner Hill Sequence, Jam Brook Complex) show no significant evidence of Laurentian-derived detritus, and our samples from what one might presume to be the lower portions of the Fredericton trough also show no evidence of Laurentian input (this presumption is based on comparisons with the findings of Dokken et al. 2018). In contrast, Silurian rocks from the eastern margin of the Central Maine basin have zircon populations suggestive of Laurentian input, similar to findings in the western portions of this same basin (Bradley et al. 2015), and the younger portions (post-Llandovery) of the Fredericton trough (Dokken et al. 2018). These findings support the hypothesis that a barrier existed between the Laurentian margin and Gondwana-derived terranes prior to about 435 Ma (Dokken et al. 2018). Alternatively, Ludman et al. (2018) suggested that little sediment input into the Central Maine basin or Fredericton trough was derived from Laurentian or Avalonian sources, but rather was largely derived from Ganderian basement that became emergent following Late Ordovician orogenesis. Clearly additional detailed sediment provenance studies in the thick, largely post-Ordovician sedimentary basins in the northern Appalachians (e.g., Connecticut Valley, Central Maine/ Aroostook-Matapedia, Fredericton) are needed to fully understand the sediment sources of these basins, the pathways for sediment transport into them, and ultimately, Silurian– Early Devonian paleogeography.