Receiver functions from the EarthScope Southeastern Suture of the Appalachian Margin Experiment broadband deployment and U.S. Transportable Array were analyzed to constrain average crustal thickness and composition across the southern Appalachians. Low Vp/Vs ratios (1.69–1.72) across the Carolina terrane and parts of the Inner Piedmont indicate that the crust has a felsic average composition. The results are consistent with models of thin-skinned thrusting of Carolina arc fragments over Laurentian basement, whereas arc collision models require significant crustal modification to explain the low Vp/Vs. New crustal thickness estimates provide constraints on the extent of the Blue Ridge crustal root. The present root may be a remnant of a broader structure formed by Alleghanian thrust loading. Root preservation is attributed to Mesozoic heating and thinning of the lower crust beneath outboard terranes, leaving colder Blue Ridge crust largely intact. However, thickened crust (50–55 km) across the region may also be inherited from continental collision during the Proterozoic Grenville orogeny.
 The Southeastern Suture of the Appalachian Margin Experiment (SESAME) is designed to study lithospheric evolution associated with two cycles of continental collision and breakup in the southeastern United States (Figure 1). During the Grenville orogeny (1.3–1.0 Ga), arc and continental collision along the eastern margin of North America (present coordinates) resulted in the formation of the supercontinent Rodinia [Hynes and Rivers, 2010]. Subsequent opening of the Iapetus ocean in the Late Proterozoic was followed by renewed plate convergence, episodic terrane accretion, and strike-slip tectonics along the North American margin during the Paleozoic [Aleinikoff et al., 1995; Hibbard et al., 2012]. Pangea formed during the Permo-Carboniferous Alleghanian orogeny [Hatcher, 2010], while continental breakup occurred in the Triassic-Jurassic [Hames et al., 2000]. In this paper, we use Ps receiver functions to evaluate the extent to which Alleghanian thin-skinned tectonics and continental rifting modified Grenville age lower crust along the Laurentian margin. The implications for crustal evolution, crustal root preservation, and lower crustal petrology are briefly discussed.
 Consortium for Continental Reflection Profiling (COCORP) studies indicate that terranes exposed in the southern Appalachians are part of a west-vergent, thin-skinned (3–15 km thick) thrust belt emplaced over the Laurentian margin during the Alleghanian orogeny [Cook and Vasudevan, 2006]. The Inner Piedmont, Carolina terrane, and Coastal Plain may be underlain by Grenville basement beneath the detachment (Figure 2) [Phinney and Roy-Chowdhury, 1989]. Alternatively, the Carolina terrane or Inner Piedmont may have been underthrust beneath the Laurentian margin in the vicinity of the Central Piedmont suture zone prior to Alleghanian collision [e.g., Anderson and Moecher, 2009; Hibbard et al., 2012].
 Previous wide-angle reflection and receiver function studies show that crustal thickness increases northwestward from ~35 km beneath the Coastal Plain to 50–56 km across the Blue Ridge Mountains [French et al., 2009; Hawman et al., 2012; Wagner et al., 2012]. The relationship between the root and present topography suggests isostatic balance, but the timing, mechanisms, and original extent of root formation remain uncertain [Hawman et al., 2012]. We present two alternative models involving Permo-Triassic and pre-Alleghanian deformation.
2 Data and Methods
 The SESAME array consists of 85 broadband seismometers deployed along three transects in Georgia, Florida, North Carolina, and Tennessee (Figure 1). In this paper, we focus on results from SESAME stations within the Carolina terrane, Inner Piedmont, Blue Ridge, and Valley and Ridge. We also incorporate data from the U.S. Transportable Array to help assess the regional significance of the SESAME results.
 For receiver function computation, we employed a frequency domain method to isolate P-to-S conversions within the crust and mantle [Langston, 1979]. For stations that exhibited strong Ps conversions and clear crustal multiples (PpPs and PpSs + PsPs), we used Zhu and Kanamori's  grid search method to determine variations in crustal thickness (H) and Vp/Vs (k) across the region. To constrain the results, we used a range of average crustal Vp values (6.2–6.6 km/s) derived from previous wide-angle experiments in the southern Appalachians [Prodehl et al., 1984; Hawman et al., 2012]. For stations that did not exhibit clear multiples, we estimated crustal thickness using Ps delay times and an assumed range of average crustal Vp and Vp/Vs values. Uncertainties range from ± 1.0 to ± 5.0 km for crustal thickness and ± 0.01 to ± 0.07 for Vp/Vs (Tables S1 and S2 in the supporting information). Methodology and error analysis are discussed in greater detail in the supporting information.
3 Crustal Structure Across the Southern Appalachians
 Stations D02–D09 within the Carolina terrane exhibit particularly strong Ps conversions and crustal multiples (Figures 1 and 2). The resulting H-k stacks from these stations provide estimates for crustal thickness of 36–37 km and average crustal Vp/Vs of 1.69–1.72 (Table S1). These results are anchored by well-constrained receiver function analyses using a Gaussian value of 5.0 for stations D02, D05, and D08 (Figure 3 and Table S1). Previous wide-angle profiling across the Carolina terrane [Hawman et al., 2012] yielded crustal thickness estimates of 37–39 km and slightly higher Vp/Vs of 1.75 ± 0.01 (see the supporting information for discussion).
 The results from the Inner Piedmont show a slight increase in Vp/Vs and a gradual increase in crustal thickness toward the Blue Ridge, in agreement with previous wide-angle results. Stations D15 and D17 yield H-k estimates of 41–43 km and 1.76–1.77 (Figure 3). Stations W29 and W31 yield similar crustal thickness estimates (40–42 km) but lower Vp/Vs ratios of 1.72–1.74. For comparison, a well-constrained H-k estimate from station GOGA (near the Inner Piedmont/Carolina terrane boundary) indicates a crustal thickness of 41–43 km and Vp/Vs of 1.72–1.76, and the average Vp/Vs across the Inner Piedmont derived from wide-angle data is 1.73 ± 0.02 [Hawman et al., 2012].
 For Inner Piedmont stations with weak multiples, crustal thickness estimates using Ps delay times and an assumed Vp/Vs of 1.76 show a gradual increase in thickness from 36 to 43 km (delay time: 4.5–5.3 s) toward the Blue Ridge (Figure 4 and Table S1). Allowing for a range of 6.2–6.6 km/s for average crustal Vp and 1.73–1.78 for average crustal Vp/Vs resulted in perturbations of 2–3 km in crustal thickness (Table S1).
 Crustal thickness estimates across the Blue Ridge and Valley and Ridge provinces show considerable variability. Assuming an average crustal Vp of 6.5 km/s based on refraction results and a Vp/Vs of 1.76 based on a well-constrained value for MYNC [Hawman et al., 2012], Ps delay times of 5.5 to 7.1 s in the Blue Ridge correspond with crustal thickness estimates between 45 and 58 km (Figure 4 and Tables S1 and S2). The location of maximum crustal thickness (~55 km) appears localized beneath stations D20, W34, W35, and W53A, in agreement with previous wide-angle results (54–56 km) and a receiver function estimate of 50–52 km at MYNC. To the northeast at V53A, crustal thickness decreases to 46 km (5.7 s), which is consistent with estimates of 46 km from the AST array [Wagner et al., 2012]. In the Tennessee Valley and Ridge, estimates of H indicate the Moho shallows to 45–48 km (5.5–6.0 s), but the crust thickens again to 54 km (6.6 s) to the northwest at station V50A along the southeast flank of the Cumberland Plateau. At stations X51A and Y51A, H-k stacks indicate thickened crust (47–50 km; 5.7–6.3 s) persists to the southwest of the Blue Ridge, despite the lower elevations. Estimated crustal thickness from FLED station FA07 (elevation: 178 m) in northern Alabama is also 50 ± 2.6 km [French et al., 2009].
 The low average crustal Vp/Vs ratios (1.69–1.72) across the Carolina terrane and parts of the Inner Piedmont indicate a felsic average crustal composition [Christensen, 1996]. The bulk composition is consistent with average crustal Vp of 6.2–6.6 km/s derived from wide-angle experiments across the region [Hawman et al., 2012]. The upper limit for average Vp (6.6 km/s) is similar to the average Vp of 6.5 km/s determined for middle-to-lower crustal felsic rocks from the exhumed Pikwitonei granulite belt in Canada [Fountain and Salisbury, 1996]. In the southern Appalachians, the low Vp/Vs ratios and low-to-moderate average Vp values (6.2–6.6 km/s) may be explained by a combination of quartzofeldspathic gneisses, metasedimentary rocks, and felsic granulites in the lower crust [e.g., Kern and Schenk, 1988]. The increase in Vp to ~7.0 km/s in the lowermost crust [Hawman et al., 2012] may be indicative of metasedimentary rocks containing garnet or sillimanite and possibly significant amounts of quartz [Fountain, 1976; van den Berg et al., 2005], rather than mafic granulites or arc rocks.
 The felsic average crustal composition across the Carolina arc terrane is consistent with thin-skinned tectonic models for the southern Appalachians, but not with lateral accretion of an intact arc complex in the vicinity of the Central Piedmont suture zone [Hibbard et al., 2012] unless significant modification of the crust occurred during or after collision. The low Vp/Vs values indicate that the crust is more silica rich than intact arc crust of andesitic-to-mafic composition [Jagoutz and Schmidt, 2012], which implies that volcanic and plutonic arc rocks exposed at the surface are not representative of lithologies at depth. Instead, arc rocks likely accreted farther outboard during the Neo-Acadian (380–355 Ma) were later transported northwestward during Alleghanian thin-skinned thrusting over either North American or Inner Piedmont basement [Anderson and Moecher, 2009]. The results are consistent with models that show that the Alleghanian detachment extends southeastward beneath the Carolina terrane and possibly the Coastal Plain [Hatcher et al., 1989].
 In the context of thin-skinned tectonics, the incorporation of Carolina arc fragments into the Appalachian orogen represents the process of crustal reworking accompanying continental collision [e.g., Ernst, 2010], rather than continental growth by lateral arc accretion [e.g., McLennan and Taylor, 1995]. The felsic composition is consistent with models of crustal recycling involving delamination of dense mafic rocks and relamination of buoyant felsic material at convergent margins [Ernst, 2010; Hacker et al., 2011]. In particular, the preservation of large volumes of felsic material may be a result of exhumation of quartzofeldspathic ultrahigh-pressure domains during continental collision (e.g., Western Gneiss Region of Norway) [Hacker et al., 2011]. The global implication is that crustal refining during convergence may be an effective means of recycling mafic rocks into the mantle and enriching the crust in felsic components [Ernst, 2010; Hacker et al., 2011]. This hypothesis is consistent with the felsic crustal structure in other orogenic domains such as the Variscides [Villaseca et al., 1999] and Irish Caledonides [van den Berg et al., 2005; Hauser et al., 2008].
 Receiver function Ps delay times indicate that a crustal root (total crustal thickness: ~55 km) is preserved beneath the high elevations of the Blue Ridge, consistent with local compensation of present topography. The correlation between topography and crustal thickness persists to the northwest, suggesting isostatic balance beneath subdued topography of the Tennessee Valley and Ridge and higher elevations beneath the Cumberland Plateau as well. Models of local and regional isostatic compensation of the Blue Ridge are generally consistent with gravity profiles and seismic constraints on crustal thickness, but the planar nature of the Appalachian detachment suggests that the middle crust has not been significantly down warped, as would be expected for both models [Hawman et al., 2012].
 We suggest instead that the present root is a remnant of a much broader region of thickened crust developed across the orogen in response to Alleghanian thrust loading.
 During Mesozoic extension, the combination of thickened crust and heating by mafic intrusions triggered thinning of the lower crust by lateral flow [McKenzie et al., 2000], allowing rebound of the Moho without significant warping of the overlying Alleghanian detachment. Extension and crustal thinning were concentrated beneath outboard terranes, leaving the crust beneath the Blue Ridge largely intact. The lower temperatures suggested by the more rigid response of Blue Ridge crust are consistent with the lack of Triassic dikes northwest of the Inner Piedmont [King, 1961]. Alternatively, given the persistence of roots for over 1 Ga [e.g., Fischer, 2002], the deep structure may be partly related to thickening inherited from Grenville continental collision. In either case, both the Blue Ridge root and thickened crust (~50 km) beneath the lower elevations in west Georgia and northern Alabama may have been preserved by retrograde metamorphic reactions in a cooling lower crust that caused an increase in lower crustal density, inhibiting uplift [Fischer, 2002; French et al., 2009].
 The low average crustal Vp/Vs from SESAME stations across the Carolina terrane and parts of the Inner Piedmont indicates a felsic average crustal composition, which is consistent with Alleghanian thin-skinned tectonics. The bulk composition suggests that crustal refining during continental collision is an effective means of enriching the crust in felsic material and recycling mafic components into the mantle. The low Vp/Vs ratios are incompatible with models of continental growth by island arc accretion along the Central Piedmont suture zone, unless removal of mafic lower crust and addition of felsic components has also occurred. The formation of the present root beneath the Blue Ridge and adjacent areas is consistent with Alleghanian collision followed by Mesozoic rifting, but the regional structure is also consistent with thickening inherited from Proterozoic Grenville tectonics.
 We thank our station hosts for their assistance and hospitality, and our field crew. Comments by two anonymous reviewers improved the manuscript. This work was funded by NSF grants EAR-0844276 (K.M.F.), EAR-0844186 (L.S.W.), and EAR-0844154 (R.B.H.).
 The Editor thanks Robert D. Hatcher Jr. and an anonymous reviewer for their assistance in evaluating this paper.