We report remanent magnetization measurements from 13 sites in Cretaceous plutonic rocks in the northern Sierra Nevada (38°N–39.5°N). By increasing the number of available paleomagnetic sites, the new data tighten constraints on the displacement history of the Sierra Nevada block and its pre-extensional position relative to interior North America. We collected samples in freshly exposed outcrops along four highway transects. The rocks include diorite, granodiorite, and tonalite with potassium-argon ages (hornblende) ranging from 100 Ma to 83 Ma. By combining our results with previous paleomagnetic determinations from the central and southern Sierra Nevada (excluding sites from the rotated southern tip east of the White Wolf–Kern Canyon fault system), we find a mean paleomagnetic pole of 70.5°N, 188.2°E, A95 = 2.6° (N = 26, Fisher concentration parameter, K = 118). Thermal demagnetization indicates that the characteristic remanence is generally unblocked in a narrow range within 35 °C of the Curie temperature of pure magnetite. Small apparent polar wander during the Cretaceous normal-polarity superchron, plus prolonged acquisition of remanence at the site level, may account for the low dispersion of virtual geomagnetic poles and relatively large K value. Tilt estimates based on overlapping sediments, stream gradients, and thermochronology of the Sierra Nevada plutons vary from 0° to 3° down to the southwest. Without tilt correction, the mean paleomagnetic pole for the Sierra Nevada is essentially coincident with the North American reference pole during the Cretaceous stillstand (125 Ma to 80 Ma). At 95% confidence, the apparent latitude shift is 1.1° ± 3.0° (positive northward), and the apparent rotation is negligible, 0.0° ± 4.7°. Correcting for each degree of tilt, which is limited to 3° on geologic evidence, increases the rotation anomaly 2.2° counterclockwise, while the apparent latitude shift remains unchanged.
Late Cretaceous plutons of the Sierra Nevada have been the target of several paleomagnetic investigations that tested for mobility of the California-Nevada margin as a consequence of transcurrent plate interactions and crustal extension. Early studies by Currie et al. (1963) and Grommé and Merrill (1965) showed Cretaceous paleomagnetic poles to be similar from the Sierra Nevada and the North America craton, albeit within large 95% confidence limits. At about the same time, Hamilton and Myers (1966) introduced the concept of a rigid Sierra Nevada block moving northwestward and rotating counterclockwise relative to interior North America as the Basin and Range Province extended several hundred kilometers (Fig. 1). Atwater (1970) applied the nascent seafloor-spreading model to explore relative motions between Pacific oceanic plates and North America and proposed that some of the relative motion had been accommodated within a broad zone extending east of the Sierra Nevada. Efforts to reduce uncertainty of the Cretaceous paleomagnetic pole position increased in the 1980s, as researchers investigated possible displacement of the Sierra Nevada (Kanter and McWilliams, 1982; Frei et al., 1984; Frei, 1986; McWilliams and Li, 1985; Li, 1988). As summarized by Frei (1986), paleomagnetic data from Kings Canyon and Yosemite National Park were consistent with northward movement and modest clockwise rotation of the Sierra Nevada. At 95% confidence, the Cretaceous pole indicated 6° ± 8° of clockwise rotation and 6° ± 6° of poleward displacement of the Sierra Nevada relative to cratonic North America. Neither measure of discordance, however, provided compelling evidence that rotation and displacement of the Sierra Nevada had been confirmed by paleomagnetic methods. Frei's (1986) analysis did confirm that the Sierra Nevada and overlying Great Valley Sequence essentially constituted a rigid block south of latitude 38°N. The other studies, as exemplified by McWilliams and Li (1985), showed that the southern tip of the Sierra Nevada Batholith, but not the main body of the mountain range west of the White Wolf–Kern Canyon fault system, had undergone substantial clockwise oroclinal bending.
More recent reconstructions of the pre-extensional position of the Sierra Nevada block are based on geologic mapping of fault offsets (Wernicke and Snow, 1998; McQuarrie and Wernicke, 2005) and plate-circuit analysis (Atwater and Stock, 1998). These models, which address the magnitude of extension in the Basin and Range Province and the space requirements at the continental margin, are generally consistent with the results of Frei (1986). However, the authors point out that Sierra paleomagnetic data describe only net rotation since ca. 90 Ma and do not constrain any mutually canceling rotations that possibly occurred during the critical period of Miocene extension. Cancellation of rotations is possible if the Sierra Nevada–North America finite rotation pole changed from the southern part of the Basin and Range Province (imparting a counterclockwise rotation) to the northern part (clockwise) during Miocene time (Atwater and Stock, 1998).
The purpose of the current study is to present additional paleomagnetic results from Cretaceous plutons north of the previous work. Our collection doubles the number of previously available sites, extending the area of coverage from latitude 38°N to 39.5°N and improving confidence in the overall result. Also, we include subsequent work that gives better confidence limits for the Cretaceous reference pole of cratonic North America (Housen et al., 2003). These improvements lead to tighter constraints on estimates of translation and rotation pertaining to the Sierra Nevada block. Finally, recent publication of paleomagnetic data from 10 Ma basalt at Sonora Peak warrants discussion in regard to sequential rotation and mobility of the Sierra Nevada during the late Miocene (Pluhar et al., 2009).
CRETACEOUS PALEOMAGNETIC REFERENCE POLES FOR NORTH AMERICA
The apparent polar wander path, which establishes a spatial reference frame for the tectonically stable part of the continent, continues to be refined for North America. Paleomagnetic compilations give a good representation of Cretaceous apparent polar wander, showing that the pole was nearly stationary relative to stable North America from 125 Ma to 80 Ma (Van Fossen and Kent, 1992; Tarduno and Smirnov, 2001; Besse and Courtillot, 2002; Housen et al., 2003; Enkin, 2006; Kent and Irving, 2010). This interval, commonly known as the Cretaceous stillstand, overlaps the age range of the Cretaceous normal-polarity superchron (ca. 118–84.0 Ma; Ogg and Smith, 2004). The superchron is essentially the same age as the magnetization lock-in period of the Sierra Nevada plutons of interest here (Table 1). Housen et al. (2003) compiled high-quality paleomagnetic data from seven rock units of Cretaceous age in the central and eastern parts of North America. Dikes, small intrusive bodies, and lavas account for six of the studies; the seventh study was composed of data from the sedimentary Niobrara Formation in Kansas, Wyoming, and Colorado. The data were grouped into 19 virtual geomagnetic poles, each representing a geologically short time period. After rejecting one outlier and using the remaining 18 virtual geomagnetic poles to calculate a grand mean, Housen et al. (2003) presented a Cretaceous (125–85 Ma) reference paleomagnetic pole of 70.1°N, 191.2°E, A95 = 2.7°. This paleomagnetic pole is similar to the pole (74.0°N, 185.4°E, A95 = 3.4°) obtained by Tarduno and Smirnov (2001) from a compilation that was limited to Cretaceous plutons (124–88 Ma) in Maine, New Hampshire, Vermont, Arkansas, and Quebec.
An alternative approach to constructing apparent polar wander paths incorporates data from several continents and compensates for plate-tectonic relative motion (Besse and Courtillot, 2002; Kent and Irving, 2010). The multicontinent approach augments the data set, allows for stricter quality selection of the data, and fills time gaps caused by absence of appropriate paleomagnetic targets on some continents. However, the approach has the disadvantage of adding the uncertainties posed by plate-reconstruction finite rotations. For the Cretaceous apparent polar wander of North America, we find a significant difference in pole positions resulting from the single continent approach of Housen et al. (2003) and the multicontinent method exemplified most recently by Kent and Irving (2010). A combination of paleomagnetic poles from Africa, South America, Australia, India, and North America for the time interval 80.5–124 Ma (poles 17–33 in table 5 of Kent and Irving, 2010) gives a reference pole in North America coordinates of 76.2°N, 190.1°E, A95 = 2.2°. The angular distance between the poles resulting from the two approaches is 6.1°, which is a relatively large difference considering the small 95% confidence limits for each reference pole.
In analyzing the Sierra Nevada paleomagnetic data for discordance, we elected to use the reference pole of Housen et al. (2003) to avoid possible bias due to the finite rotations used in plate-tectonic reconstruction. We note, however, that the multicontinent reference pole implies an expected declination at the Sierra Nevada location directed 8° (±4° at 95% confidence) more to the east than is predicted from the pole of Housen et al. (2003).
PALEOMAGNETIC SAMPLING AND LABORATORY METHODS
In order to complement the work of Frei et al. (1984) and Frei (1986), we sampled Cretaceous granitic rocks along a series of four highway transects across the northern sierra Nevada north of Yosemite National Park (Fig. 1). Following the experience reported by Currie et al. (1963), Grommé and Merrill (1965), and Frei (1986), we attempted as far as possible to sample only excavated highway cuts, and also to concentrate on the cognate mafic inclusions (Fig. 2) that are abundant throughout the Sierra Nevada granitic rocks (Pabst, 1928; Dodge and Kistler, 1990). In addition, we attempted to work in areas for which published geologic mapping was available. These criteria restricted our sampling to a considerable degree. Only one unexcavated exposure was sampled; this glaciated outcrop yielded random and unstable natural remanent magnetization (NRM) directions and will not be discussed further. At four of the remaining 14 sites (4H528, 4H577, 4H585, and 4H610), no mafic inclusions were present. Overall, our data were obtained from 67 specimens of mafic inclusions and 61 specimens of granitic rock that was host to inclusions at 10 sites and the only rock present at the rest. In Table 1, we list the site numbers or names from our study and previous studies (mostly shown on Fig. 1), the names of the sampled plutons where available, and all of the available radiometric ages, both potassium-argon (K-Ar) and uranium-lead (U-Pb). The total range in age of the Cretaceous plutons for which paleomagnetic directions have been published is from 100 Ma to 74 Ma, while the range of ages of the plutons used in our analysis is 100 Ma to 83 Ma.
Sampling was done with a portable gasoline-powered core drill, and separation between cores along the outcrop was kept to a minimum of 3 m wherever possible. Core orientation was by magnetic compass using a stage on a 1-m-long tube, so that the compass was distant from any excessively magnetic rock. Magnetic bearings were corrected to true geographic azimuths by back-sighting from the compass platform to landmarks, and no evidence of lightning strikes was found at any of the sampled highway cuts. The customary 2.46-cm-diameter cores were sliced into specimens that were 2.27 cm long.
Magnetic measurements were made with a three-axis superconducting magnetometer, and alternating-field (AF) demagnetization was done with a commercial 400 Hz shielded two-axis tumbling device. Thermal demagnetization of selected specimens was done with a noncommercial oven in air and in a field less than 5 nT. As in all previous studies of Sierra Nevada plutonic rocks, AF demagnetization was the method of choice for magnetic cleaning. For each site, two or three specimens were subjected to detailed pilot AF demagnetization, and, based on these results, one or two “blanket” AF cleaning levels were selected for the entire site. For the six sites for which two cleaning levels were tried, the angular standard deviation remained above 10°; dispersion increased after the second level for four of these sites, remained constant for one, and decreased for only one. For two more sites, it was not possible to select an optimum AF demagnetizing field, so all specimens were subjected to progressive demagnetization, and the characteristic magnetization was obtained by fitting least-squares lines to each demagnetization path using the method of Kirschvink (1980). The average percentage decrease in angular standard deviation after magnetic cleaning was 28%, with a range from 67% to −29%; in other words, the specimens from most sites responded fairly well to AF cleaning, but at two of the 14 sites, the angular standard deviation increased.
We measured bulk magnetic susceptibility of all specimens that were used in the alternating field treatments. The susceptibility meter operated with a 0.1 mT peak alternating field of 800 Hz, and the maximum sensitivity of the instrument was 1.2 × 10−6 (SI unit system).
The intensity of NRM is approximately the same in both the mafic inclusions and in the host granitic rock; in the inclusions, the geometric mean is 0.33 A/m, with maximum and minimum at 4.0 and 0.009 A/m, respectively, while in the host granitic rock, the geometric mean is 0.27 A/m, with maximum and minimum at 2.0 and 0.01 A/m, respectively (Fig. 3A). In the host granitic rock, a ceiling value for NRM intensity less than 1 A/m exists for all but one specimen; this ceiling does not exist for the mafic inclusions. Magnetic susceptibilities are similar for the host granitic rock and mafic inclusions; the geometric mean susceptibility for all specimens used in the AF demagnetization experiments is 0.02 (Fig. 3B).
A convenient measure of the magnetic “hardness” or stability of NRM is the median destructive field (MDF), defined as the strength of the alternating field required to reduce the NRM intensity in a specimen by one half. We compared the NRM intensity (normalized to the magnetic susceptibility) with the MDF for each demagnetized specimen (Fig. 3C). In both kinds of rock, higher MDF values tend to be associated with higher NRM/susceptibility ratios. Little difference exists between the distributions of MDF values between mafic inclusions and enclosing rock; in other words, the mafic inclusions are not significantly harder magnetically than the enclosing rock, an unexpected result. Likewise, we find no correlation, negative or otherwise, between angular standard deviation and percentage of mafic inclusion material for the 14 sites.
The demagnetization characteristics of two selected specimens representing contrasting stabilities are shown in Figures 4 and 5. Figure 4 shows AF and thermal demagnetization of one of the typically stable specimens, from a fine-grained mafic inclusion. Except for a negligible component that vanishes at the first AF or thermal step, only one component of magnetization existed in this specimen; stated another way, the NRM is almost exactly the characteristic remanent magnetization. The narrow distribution of unblocking temperatures is remarkable and has not previously been reported for Sierra Nevada granitic rocks, although Frei et al. (1984) showed one example from the Mount Gibson granodiorite that was nearly as stably magnetized. Demagnetization behavior of a typical partly stable specimen is shown in Figure 5 and resembles the vector AF demagnetization diagram shown by Frei (1986). These “noisy vector component diagrams” were interpreted by Frei (1986) as representing removal of a single component of magnetization and the addition at each of the steps of a spurious component of magnetization with random orientations and magnitudes. This interpretation is borne out by comparison of the AF and thermal demagnetizations illustrated in Figure 5, and also by petrographic observations described later herein. In spite of the noisiness of the thermal demagnetization evident in the vector component diagram, the narrowness of the main unblocking temperature distribution is obvious in Figure 5.
In total, 29 specimens were thermally demagnetized, 15 from mafic inclusions and 14 from host granitic rock. The results are well represented by the examples in Figures 4 and 5. It is convenient to define a “peak –dJ/dT temperature” as the temperature at the midpoint of the steepest segment of the thermal demagnetization diagrams. For both the mafic inclusions and the host rock, this temperature was 569 °C, with standard deviations (s.d.) of 14° and 6 °C, respectively. The widths of the major unblocking temperature distributions were 33° ± 15 °C (s.d.) for the mafic inclusions, and 35° ± 11 °C (s.d.) for the host rock. In our thermal demagnetization equipment, the thermocouples tend to reach slightly different (usually higher) temperatures than the interiors of the specimens, and the reported standard deviations include this effect. We interpret the thermal demagnetization data as representing no Curie temperature higher than that of pure magnetite (580 °C), and moreover there is no evidence of material with a lower Curie temperature, such as titanomagnetite. This interpretation is supported by the published X-ray diffraction and microprobe data for these rocks, which report virtual absence of Ti from magnetite (Currie et al., 1963; Grommé and Merrill, 1965; Ague and Brimhall, 1988).
Magnetite in Sierra Nevada granitic rocks commonly occurs as subhedral grains ranging in diameter from ∼250 μm down to ∼5 μm, and it is often partly or completely surrounded by sphene. This association is significant in that it represents subsolidus equilibration of original titanomagnetite resulting in loss of Ti, presumably continuously during slow cooling of the batholithic rocks (Grommé and Merrill, 1965; Ague and Brimhall, 1988). Magnetite that has been so purified and annealed in nature would be expected to be magnetically soft, resulting in Koenigsberger ratios (Q) less than unity, as have been observed (Currie et al., 1963; Grommé and Merrill, 1965). Thus, it is easy to account for the noisiness of many of the vector component diagrams by blaming the largest magnetite grains for easily acquiring “spurious,” randomly directed components of isothermal remanent magnetization in the laboratory. Parenthetically, we note that the results reported herein were obtained without benefit of a magnetically shielded enclosure.
It is more of a problem to account for the stability of the NRM in many of our specimens (Fig. 4). We examined a small but representative set (10) of polished thin sections in both reflected and transmitted light, to look for other magnetite, and this has been observed. In these rocks, the hornblendes are usually clear and do not contain inclusions. In some specimens of mafic inclusions, however, the hornblende grains have dark centers. At high magnification (600× or more), these dark centers are seen to be aggregates of isotropic reflective grains, obviously magnetite (Fig. 6A). These magnetite inclusions are elongate with aspect ratios around 5, are typically 3 μm long by 0.5 μm wide, and appear to be needles rather than platelets (Fig. 6B). They occur in two orientations, perfectly uniform and orthogonal within single hornblende grains. The most common orientation is at a small angle to the prismatic or z axis but very close to the optical extinction direction. Closer examination of hornblende-hosted magnetite inclusions with a scanning electron microscope showed elongate forms ∼2 μm wide with length/width ratios up to 10 (Figs. 6C and 6D). These elongate magnetite inclusions are too large to have single-domain magnetic structure (Butler and Banerjee, 1975; Frandsen et al., 2004; Ozdemir and Dunlop, 2006), but they are small enough to be termed pseudo–single-domain (for summary, see Dunlop, 1990) and probably account at least in part for the stability of much of the NRM and the narrowness of the unblocking temperature distributions.
The in situ paleomagnetic data for the 14 new sites in the northern Sierra Nevada are shown in Table 2 and are illustrated on Figure 7A. For all sites, each 95% confidence circle includes neither the axial dipole field direction nor the present-day field direction. One site (4H610) has a distinctly aberrant mean magnetization direction; although the outcrop is very large, it is in a topographic low and is surrounded by vegetation. Because, unlike all the others, this site is not so clearly connected to the body of the Sierra Nevada Batholith, we suppose that it has been displaced by landsliding and do not consider it further.
We also list in Table 2 the published in situ paleomagnetic data for three sites in the Grace Meadow and Mount Gibson plutons in the northern part of Yosemite National Park (Frei et al., 1984) and for nine sites in the Kings Canyon area in the south-central Sierra Nevada (Frei, 1986). In the case of the Kings Canyon area data, we used only the data that showed no evidence of population elongation or “streaking” as given by Frei (1986). For these previously published data, the virtual geomagnetic poles for each site were recalculated assuming that the plutons have not been tilted at all since becoming magnetized. Following Christensen (1966), Frei et al. (1984) and Frei (1986) had applied a correction for an estimated post-Eocene westward tilt of 1° as estimated from the present slopes of the Eocene auriferous gravels and the later Cenozoic volcanic rocks. The reason we did not follow this practice is that considerable disagreement remains concerning the amount of tilt, as derived from several lines of evidence. Unruh (1991) determined that Oligocene, Miocene, and Pliocene deposits in northern Sierra Nevada canyons dip 1.2°–1.6° southwest. The increasing vertical gradient with age was inferred as evidence for progressive tilting of the plutonic basement rocks with uplift of the eastern side of the Sierra Nevada block. In contrast, new stable isotopic and other geochemical paleoaltimetry data from overlying early and middle Cenozoic rocks (the so-called Superjacent Series) have resulted in a significant revision of the uplift history of the Sierra Nevada (Henry, 2009; Best et al., 2009; Cassel et al., 2009a, 2009b, and references therein) and are consistent with a tilt estimate near zero. The implication of these new data is that the late Eocene auriferous gravels and the overlying Oligocene rhyolitic ash-flow tuffs and andesitic tuff-breccia were at nearly their present elevations upon emplacement. Nearly all of the present elevation of the mountain range had been achieved continuously, starting with the solidification and initial cooling of the batholithic rocks; therefore, the dips of the Superjacent Series reflect depositional gradients rather than tilt.
The maximum tilt estimate comes from tracer thermochronology by McPhillips and Brandon (2010), who inferred post–35 Ma tilt of the plutonic bedrock to be 3.4° ± 0.8° down to the southwest in the Kings River and San Joaquin River drainages. We interpret this tilt estimate as the upper limit, because it exceeds the dip of Oligocene superjacent volcanics in the northern Sierra by 1° to 2°. We considered rotation calculations resulting from the range of tilt corrections varying from 0° to 3°. Essentially, correcting for each degree of tilt about strike 150° reduces the mean declination by ∼2.2°; the accompanying changes in inclination are negligible. The discussion in the remaining part of this section assumes no tilt correction has been applied.
All of the data of the upper part of Table 2 are illustrated in Figure 7B, with the overall mean direction as given in Table 2. The site-mean directions taken together do not appear to have a circular Fisher (1953) distribution; instead, the directions tend to be spread in inclination along the paleomagnetic meridian. We applied principal component analysis to the directions and the corresponding virtual geomagnetic poles to quantify the elongation of both kinds of data, following the eigenvalue method discussed by Tauxe and Kent (2004). For the directions, the elongation ratio is 2.4, and the azimuth of the minimum eigenvector is 14° counterclockwise relative to the paleomagnetic meridian. For the virtual geomagnetic poles, the ratio is 1.5, and the azimuth of the minimum eigenvector is 35° clockwise relative to the paleomagnetic meridian. We note that although neither of these ratios is statistically significant at 95% confidence for this relatively small sample set with N = 26 (critical value = 2.64; Schmidt, 1990), the elongation of virtual geomagnetic pole is generally consistent with the variation that would be expected at sites spanning three degrees of latitude within the Sierra Nevada. In Figure 8, the measured paleomagnetic colatitudes of virtual geomagnetic poles from the Sierra Nevada plutons are compared with the predicted Cretaceous colatitude (dashed line) derived from the geocentric axial dipole assumption.
At this point, we account for the data in Table 2 that we omitted from the mean. We dealt with our site 4H610 in the foregoing. Our evidence for omitting the others is illustrated in Figures 8 and 9. Referring to Figure 1, sites BVS (Kanter and McWilliams, 1982), KELSO, and EDMON (McWilliams and Li, 1985; Li, 1988) are all southeast of the White Wolf–Breckenridge–Kern Canyon fault system. Their discordance is mainly in paleomagnetic declination (Fig. 9) and represents a well-established progressive clockwise structural bend of the southernmost Sierra Nevada, mainly in plan (Kanter and McWilliams, 1982; McWilliams and Li, 1985). More surprising values are the discordances of the paleomagnetic directions in the large Mount Givens pluton (MTG) (Gilder and McNulty, 1999) and the Lake Edison pluton to the east (LKE) (Ross, 1988). The Mount Givens data are discordant only in paleomagnetic declination, but the Lake Edison data are discordant in both paleolatitude and declination. By comparing the observed direction in each pluton at their respective mean site locations with the corresponding directions derived from the North American reference pole of Housen et al. (2003), we obtain the following rotation parameters required to bring the observed directions into parallelism with the reference: For the Mount Givens pluton, the minimum rotation is 11° counterclockwise around an axis plunging 24° and trending 157°; the equivalent dip and strike are 12°SW and 158°, respectively. For the Lake Edison pluton, the minimum rotation is 3° counterclockwise around an axis plunging 6° and trending 233°; the equivalent dip and strike are 3°NW and 132°, respectively. If we compare the observed directions with the mean given in Table 2, we find mostly similar rotation parameters: For the Mount Givens pluton, the minimum rotation is 11° counterclockwise around an axis plunging 25° and trending 161°; the equivalent dip and strike are 11°W and 162°, respectively. For the Lake Edison pluton, the minimum rotation is 10° counterclockwise around an axis plunging 19° and trending 197°; the equivalent dip and strike are 10°W and 194°, respectively. No matter how the calculations are performed, we are forced to conclude that the postmagnetization tilts in the Mount Givens and Lake Edison plutons are significant in every respect, and moreover the paleomagnetic directions in the two are discordant to one another. Ross (1988) did not make any comparison with a reference pole, but Gilder and McNulty (1999) compared their mean direction with published Sierra Nevada paleomagnetic data from Cretaceous plutons and with the composite, multicontinent North American reference pole of Besse and Courtillot (2002). They accounted for the tilts of both plutons by rotation during uplift that was accommodated by mapped marginal ductile shear zones. Gilder and McNulty (1999, p. 921) estimated for the Mount Givens pluton a tilt of ∼10° westward around an axis (320°) “that parallels the dominant regional fabric,” similar to what we have reported here.
The magnetizations in these plutons were evidently acquired over relatively long periods of time that almost certainly overlapped, an argument that was briefly developed by Frei et al. (1984) and Frei (1986), and that is substantiated by the petrographic observations of magnetite in these plutons by us and by Grommé and Merrill (1965) and Ague and Brimhall (1988). Lock-in time of the remanent magnetization depends on the cooling rate at depth and the magnetite blocking temperature range. Dumitru (1990) compiled thermochronologic data for Cretaceous plutonic rocks in the areas of Yosemite Valley and Kings River Canyon to derive a generalized time-temperature profile. The plutons cooled from crystallization temperatures to 350 °C within a few million years just prior to ca. 85 Ma, as indicated by the differences between U-Pb and K-Ar ages. We estimate that cooling rates were ∼100–200 °C/m.y. during this time period. By extrapolation of the laboratory unblocking temperature range (545−580 °C) to natural cooling conditions, we surmise that the characteristic remanent magnetization was acquired as the plutons cooled from 580 °C to ∼500 °C (Pullaiah et al., 1975; Dunlop and Ozdemir, 1997). Therefore, it is likely that the characteristic magnetizations at the site level and perhaps specimen level were acquired over several hundred thousand years. Prolonged acquisition of remanent magnetization is expected to reduce angular dispersion of the site mean directions due to long-term averaging of geomagnetic secular variation. Also, the Cretaceous normal-polarity superchron (84–118 Ma) was an unusual period of reduced apparent polar wander, with few if any polarity reversals; angular dispersion of virtual geomagnetic poles during this interval was reduced compared to periods with higher reversal rates (McFadden et al., 1991; Cronin et al., 2001; Tarduno et al., 2002; Biggin et al., 2008). Relatively slow acquisition of remanence during the Cretaceous period of unusually stable geomagnetic field behavior may account for the small angular standard deviation (S = 7.4°; 95% confidence interval, 6.2°–9.2°; Cox, 1969) that we observe for the mean virtual geomagnetic pole. The measured dispersion is substantially less than the global-compilation value (S = 12.2°, 10.1°–15.4°) for latitude 38°N during 80–110 Ma (McFadden et al., 1991).
Comparison of the in situ Sierra Nevada mean data with the corresponding reference paleomagnetic pole (Housen et al., 2003) for the North American plate shows that the two poles are essentially coincident (Fig. 10). Applying the method of Debiche and Watson (1995), we find at 95% confidence that the apparent latitude shift of the Sierra Nevada is 1.1° ± 3.0° and that the apparent rotation is 0.0° ± 4.7°; in other words, neither is significantly different from zero. Correcting for tilt of 2° down to the southwest gives an apparent latitude shift of 1.4° ± 3.1° (northward) and apparent rotation of −4.4° ± 4.6° (counterclockwise), i.e., also insignificant. Apparent counterclockwise rotation increases to −6.6° ± 4.6° if tilt was as much as 3°, which we consider to be the upper limit of tilting.
Uncertainties posed by alternative reference pole compilations also affect the calculation of declination discordance for the Sierra Nevada. All options favor counterclockwise rotation. The mid-Cretaceous North American pole of Tarduno and Smirnov (2001), which was derived exclusively from plutonic rocks, gives a declination discordance of −4.3° ± 4.8° and a latitude discordance of −2.1° ± 3.3°. The 124–80.5 Ma reference pole that was derived by Kent and Irving (2010) under the multicontinent approach implies declination discordance of −7.8° ± 4.1° and latitude discordance of −2.0° ± 2.7°. Among these options, the only significant determination of discordance is given by the reference pole of Kent and Irving (2010). Their pole indicates counterclockwise rotation of 8° for the Sierra Nevada block.
Paleomagnetic studies from Cenozoic rocks would be useful to constrain tectonic displacement of the Sierra Nevada block during the critical period of Basin and Range extension. Recently, results were obtained from the volcanic Table Mountain Formation of the Stanislaus Group of Miocene age. Pluhar et al. (2009) reported paleomagnetic sampling of the latite, which yielded an age of ca. 10.4 Ma (40Ar/39Ar) from a sequence of 23 flows near Sonora Peak (Fig. 1). No tilt correction was applied to the Sonora Peak section under the assumption that the small westward dip of the flows followed an original depositional gradient. Magnetization directions from the 400-m-thick section are dominantly of normal polarity, but they include two reversed-polarity events. Angular dispersion of the flow virtual geomagnetic pole is compatible with long-term records of secular variation at the site latitude, from which we infer that the mean virtual geomagnetic pole adequately represents the axial dipole field. The Table Mountain Formation mean virtual geomagnetic pole is 83.3°N, 101.5°E (A95 = 5.3°, angular standard deviation = 14.0°). Using the Miocene (16 Ma) Columbia River Basalt reference pole of Mankinen et al. (1987), we calculate apparent rotation of −3.7° ± 6.4° (counterclockwise) and a latitude anomaly of 5.6° ± 5.2° (northward). Significant vertical-axis rotation of the Sonora Peak area after 10 Ma is not discernible from this paleomagnetic sampling; the apparent mean latitude shift is 4° more northward than the value we obtained from the Cretaceous plutons. The range of northward latitude shift compatible with both confidence limits of the Cretaceous and Miocene paleomagnetic poles is 0.4° to 4.1°.
Our analysis of the plutonic and younger flow data is relevant to tectonic models that describe the Cenozoic extensional displacement of the Sierra Nevada block. Wernicke and Snow (1998) proposed a dual rotation model in which the Sierra underwent: (1) counterclockwise rotation prior to 15 Ma, reflecting extension in the northern Basin and Range, and (2) clockwise rotation of 20° beginning 8–10 Ma, to account for transtensional displacement across the Death Valley region. This model predicts net rotation of the Cretaceous batholith to be reduced to the range 2° counterclockwise to 9° clockwise, which is compatible with the Sierra Cretaceous paleomagnetic pole. However, paleomagnetic data from the Miocene (ca. 10.4 Ma) Table Mountain Formation at Sonora Peak contradict the large clockwise rotation in the second stage of the model. An alternative reconstruction, as proposed by Atwater and Stock (1998), calls for a modest 5° of counterclockwise rotation of the Sierra–Great Valley block in Neogene time. Their model fits within the paleomagnetic constraints presented here. We note that the current rotation rate for a central point in the Sierra Nevada microplate is ∼0.3°/m.y. counterclockwise, as determined from the global positioning system (GPS)–derived rotation pole of Dixon et al. (2000) (see also Argus and Gordon, 1991). If we assume that the current rotation rate has been in effect since 10 Ma, when the most active phase of extension began, the implied vertical-axis rotation (3° counterclockwise) is permissible within the 95% confidence limits for the Miocene and Cretaceous paleomagnetic poles.
In regard to the latitude shift of the Sierra Nevada relative to the Colorado Plateau, the reconstruction of Wernicke and Snow (1998) called for 250–300 km of WNW displacement since 16 Ma. The predicted latitude shift corresponding to this range of displacement vectors is 1.1° to 1.4° northward, within the intersecting range (0.4° to 4.1°) of 95% confidence limits for the Cretaceous and Miocene paleolatitude discordances.
SUMMARY AND CONCLUSIONS
Road cuts in Cretaceous plutonic rocks of the Sierra Nevada yield stable, well-defined magnetization directions, which show very little difference between the mean virtual geomagnetic pole for the Sierra Nevada block as a whole and the Cretaceous reference pole for interior North America. We have extended sampling of the Sierra Nevada to the latitude band between 38°N and 39.5°N, and we have doubled the number of available paleomagnetic sites in the Cretaceous batholith. Several areas within the larger batholith complex reveal discordant paleomagnetic poles. The White Wolf–Kern Canyon fault system is a fundamental boundary that separates the significantly rotated southern tip of the batholith from the main body to the northwest (McWilliams and Li, 1985). Although the Kern Canyon fault shows relatively recent activation, the profound unroofing and deformation of the southern terrane along this boundary likely occurred in Late Cretaceous and Paleocene time (Wood and Saleeby, 1997; Saleeby et al., 2009). Localized rotations have occurred near ductile shear zones within two Cretaceous plutons, according to paleomagnetic results from the Mount Givens and Lake Edison areas (Ross, 1988; Gilder and McNulty, 1999).
Granitic rocks of the Sierra Nevada contain abundant magnetite, the product of subsolidus equilibration of titanomagnetite resulting in loss of titanium during slow cooling of the batholith. We observed high stability of natural remanent magnetization, which may be linked to pseudo–single-domain grains occurring as magnetite inclusions within hornblende and other mafic minerals. Thermal demagnetization of representative specimens revealed narrow unblocking temperature distributions, typically within 35 °C of the Curie temperature of magnetite (580 °C). Extrapolated to natural cooling conditions, acquisition of the characteristic remanent magnetization occurred as the plutons cooled from 580 °C to ∼500 °C. Cooling history, as presented by Dumitru (1990) from zircon and K-Ar dating, leads us to conclude that the within-site magnetization in these plutons represents several hundred thousand years of geomagnetic field activity. Long-term averaging of secular variation and extremely slow apparent polar wander of the Cretaceous normal-polarity superchron (118–84 Ma) explain the relatively low angular standard deviation (S = 7.4°) of virtual geomagnetic pole obtained in paleomagnetic studies of the Sierra Nevada.
Without tilt correction, magnetization directions from the Cretaceous Sierra Nevada batholith yield a mean paleomagnetic pole that is essentially indistinguishable from the cratonic North American reference pole of Housen et al. (2003). The data cover a large part of the Sierra Nevada block, spanning an area roughly 50 km × 375 km. The vertical-axis rotation determination is 0.0° ± 4.7°, and the apparent latitude shift is 1.1° ± 3.0° (positive northward). Tilt estimates vary from no tilt to 3° down to the southwest for the broad western slope of the Sierra Nevada. The maximum tilt determination, which is based on thermochronology of plutonic rocks in the central Sierra Nevada, is estimated at 95% confidence to be 3.4° ± 0.8° (McPhillips and Brandon, 2010). Each degree of tilt correction reduces the mean paleomagnetic declination by 2.2°, so that the estimated rotation in the counterclockwise sense reaches significance at 95% confidence once the tilt correction exceeds 2.2°.
To date, a single paleomagnetic pole has been determined from Miocene volcanic flows that overlie granitic rocks in the central Sierra Nevada. Twenty-three flows from the Table Mountain Formation, dated at ca. 10.4 Ma, provide an adequate sampling of secular variation to represent the axial dipole field (Pluhar et al., 2009). As with the in situ Cretaceous pole, the Miocene data present no evidence of a significant declination anomaly (−3.7° ± 6.4°, counterclockwise). The inferred latitude anomaly from the Miocene paleomagnetic pole is 5.6° ± 5.2° (northward).
Considered together, the Cretaceous and Miocene paleomagnetic poles from the Sierra Nevada tighten constraints on tectonic reconstructions that restore Neogene extension across the Basin and Range Province. The model of Atwater and Stock (1998), which calls for modest (5°) counterclockwise rotation of the Sierra Nevada block, is permissible within the 95% confidence limits of both poles. A latitude shift of ∼1° northward, as determined in the geologic reconstruction of Wernicke and Snow (1998), fits the paleolatitude data from the Sierra Nevada block.
We wish to thank and acknowledge the following people for their assistance during this project. Victoria Pease (currently at Stockholm University) assisted during the field work. Dave S. Parks performed most of the remanent magnetization measurements in the U.S. Geological Survey (USGS) laboratory. Scanning electron microscopy was provided by Robert Oscarson (USGS). Helpful reviews by Stuart Gilder, Jon Hagstrum, and Robert Simpson are much appreciated.