Abstract

We report paleomagnetic data for the Lower to Middle Jurassic La Silla and Todos Santos formations of southern Mexico, in west-central Chiapas and the Tehuantepec Isthmus region. Volcanic rocks and red beds of these formations were deposited prior to or during the early stages of Gulf of Mexico opening. Dual-polarity characteristic magnetizations reside primarily in hematite and pass intraformational conglomerate, regional tilt, and reversal tests; they are thus interpreted as primary magnetizations. Our sampling sites are concentrated in three localities; around La Angostura Lake, 17 accepted sites yield a tilt corrected mean of declination (Dec) = 325°, inclination (Inc) = 4.6° (k = 11.9, α95 = 10.8°); in the Matías Romero region, the mean is Dec = 312.9°, Inc = 3.2° (based on only seven sites); and in the Custepec area, Jurassic andesitic dikes intruding rocks of the Permian Chiapas Massif yield a corrected mean of Dec = 335.0°, Inc = 5.0° (six sites). The mean directions are discordant with respect to expected North America reference directions, and indicate a counterclockwise rotation of 35° to 40°. Inclinations indicate deposition or emplacement at near equatorial paleolatitudes (2.1°N ± 3.4°). This paleolatitude estimate is statistically indistinguishable from those previously observed in the La Boca Formation of the Huizachal Group in northeast Mexico. The localities we sampled in southern Mexico are separated by ∼150 km, suggesting that the paleomagnetic signature of these rocks reflects regional-scale rather than local deformation. These Jurassic paleomagnetic directions support a rotational origin for the Gulf of Mexico. The data are also consistent with an Early to Middle Jurassic reconstruction that places the Chiapas Massif offshore the Tamaulipas state in the western Gulf of Mexico, and an Euler rotation pole for the Maya Block for this time period in the eastern gulf. The apparent polar wander path defined by paleomagnetic poles for the Chiapas Massif and Jurassic rocks reported here suggests that relative motion between North America and the Maya Block occurred between Late Permian and Early Jurassic time, during a protracted rifting phase, and then in the Late Jurassic in association with seafloor formation in the Gulf.

INTRODUCTION AND PREVIOUS WORK

It is generally accepted that the Gulf of Mexico formed by counterclockwise rotation of the Maya (or Yucatán) Block in the Late Jurassic following a protracted episode of continental rifting that initiated in the Late Triassic (Pindell and Dewey, 1982; Hall et al., 1982; Pindell, 1985; Buffler and Sawyer, 1985; Pindell et al., 2006). Independent motion of the Maya Block with respect to both North and South America is supported by the development of rift basins as well as Upper Jurassic passive margin sequences surrounding it. The inferred timing of counterclockwise rotation in the Late Jurassic is constrained by (1) plate kinematics (Pindell and Dewey, 1982; Pindell et al., 1988; Bird et al., 2005; Pindell et al., 2006); (2) the Jurassic stratigraphy of the Gulf and northeast Mexico (e.g., Winker and Buffler, 1988; Salvador, 1987; Goldhammer, 1999), as it is generally assumed that rotation of Yucatán post-dates Callovian salt deposition; (3) isotopic (190.4 ± 3.4 Ma) age determinations of rift-related diabase dikes offshore of Yucatan (Schlager et al., 1984); and (4) paleomagnetic data for the San Ricardo Formation along the northeast flank of the Chiapas Massif (Guerrero-García et al., 1990). The rotation of the Maya Block itself is supported by paleomagnetic results for the Late Permian Chiapas Massif (Molina-Garza et al., 1992) as well as the tectonic grain in the deep Gulf of Mexico (Scott and Peel, 2001). The granitoids of the Chiapas Massif record west-directed characteristic magnetizations that yield a paleomagnetic pole position in the central equatorial Pacific. These data have been interpreted to reflect large magnitude counterclockwise rotation of the Maya Block with respect to North America (Molina-Garza et al., 1992). The rotation indicated by paleomagnetic data for the Chiapas Massif, however, is larger (∼70°) than suggested in plate reconstructions (35° –55°).

The basement structure of the western Gulf of Mexico is thought to be dominated by attenuated crust whose primary features were defined by the south-southeast drift of the Maya Block along the eastern margin of Mexico (Buffler, 1983; Buffler and Sawyer, 1985; Ewing, 1991; Marton and Buffler, 1994; Wawrzyniec et al., 2003; Wawrzyniec et al., 2004; Ambrose et al., 2003). This is supported by a steep north-south–trending basement step and associated geophysical anomalies, including a coast-parallel magnetic maximum and a steep gravity gradient offshore of the Mexican states of Tamaulipas and Veracruz. These anomalies appear to follow the trace of the Tamaulipas-Golden Lane-Chiapas transform (Pindell, 1985), a structure considered to have accommodated rotation of the Maya Block. It also has been proposed that the Maya Block was rotated about an Euler pole in the eastern Gulf region, located near the southern Florida peninsula or western Cuba (e.g., Pindell and Dewey, 1982; Hall and Najmuddin, 1984; Marton and Buffler, 1994; Pindell et al., 2006).

Because of uncertainty in the magnitude of rotation, several details in the reconstruction of the Maya Block in the northern Gulf region remain unsolved, as illustrated in Figure 1. For example, Ross and Scotese (1988) reconstruct the Maya Block farther west than other models. Pindell and Kennan (2009), on the other hand, propose a tighter fit, and the Maya Block is placed farther north than other models, while Dickinson and Lawton (2001) locate the Maya Block farther east than other models. Similarly, the orientation of the rift structures in continental lithosphere that facilitated opening of the Gulf is uncertain, and somewhat controversial (Pindell et al., 2006; Exxon, 1985; Salvador, 1987). Some authors have suggested that relative motion between the Chiapas Massif and the Maya Block occurred during the rifting process or at a later time (e.g., Ross and Scotese, 1988; Dickinson and Lawton, 2001). Therefore paleomagnetic data of Permian age from the Permian Massif may not reflect with sufficient accuracy the rotation of the Maya Block during opening of the Gulf of Mexico. Furthermore, plutonic rocks lack reference to paleohorizontal, and, in the absence of robust field relations involving overlying stratified rocks, paleomagnetic data from them are inherently of lower reliability than those from stratified rocks. The limited paleomagnetic data for the Todos Santos Formation reported by Molina-Garza et al. (1992) suggest that Jurassic rotation of the Maya Block is closer to the 35° –45° estimate that has been recently proposed in several paleogeographic reconstructions (e.g., Mickus et al., 2009; Pindell et al., 2006), but the overall quality of that Todos Santos paleomagnetic data set from the Tehuantepec region is inadequate for a reliable reconstruction. The paleopole for the Chiapas Massif, and the larger rotation inferred by Molina-Garza et al. (1992), may thus reflect internal deformation of the Maya Block during the rifting phase of continental breakup, or other tectonic processes such as Neogene deformation.

Other paleomagnetic data suggest a more complex scenario for Mesozoic deformation (Steiner, 2005). Paleomagnetic data from the Santa Rosa Group in the Maya Mountains have been interpreted by Steiner (2005) to indicate ∼180° of rotation of the Maya Block since the mid-Permian. In essence, paleomagnetic data from the Maya Mountains are similar to results from the Chiapas Massif. In the Maya Mountains, early Paleozoic intrusions that have been affected by a late Paleozoic thermotectonic event and yield Permian–Triassic K-Ar dates yield a paleomagnetic pole in the equatorial Pacific near the pole for the Chiapas Massif. Mid-Permian sedimentary rocks of the Santa Rosa Formation yield shallow, south-southwest magnetizations defining a pole at 62.5°N/22.6°E. The Santa Rosa Formation paleopole for the Maya Mountains is at an angular distance of 27° ± 17° from the pole reported for the upper Paleozoic sequence from the Chicomuselo uplift in Chiapas (the pole is located at 74.2°N–95.4°E; Gose and Sánchez-Barreda, 1981). Both poles fall near the Jurassic segment of the North American apparent polar wander path. Molina-Garza et al. (1992) reported similar, shallow and south-directed, magnetizations for Todos Santos strata in the Motozintla area of the Chicomuselo uplift, near the point where the Polochic fault system crosses the Mexico-Guatemala border. They thus interpreted the shallow south-directed magnetizations in Permian strata to be secondary, reflecting a time period of remagnetization in the Jurassic. Steiner (2005) argued, however, for a primary, Permian age of the magnetization of the Santa Rosa Formation, and thus a larger rotation than inferred by Molina-Garza et al. (1992). There is, however, relatively good agreement between an Early Triassic paleopole for the Maya Mountains and a Permian–Triassic paleopole for the Chiapas Massif.

In order to better understand the kinematics of the Maya Block during the opening of the Gulf of Mexico, we collected paleomagnetic data from the Todos Santos Formation in central Chiapas and the Tehuantepec Isthmus region. Additional data were collected from the La Silla Formation, a recently recognized sequence of volcanic rocks that underlies Todos Santos strata (Godínez-Urban et al., 2011), and from associated Jurassic dikes in the Chiapas Massif. These data, together with a reappraisal of the relevant circum-Gulf Jurassic geology, allow us to better define the rotation parameters (the Euler pole and amount of associated rotation, with error estimate) of the Maya Block for the Jurassic.

REGIONAL GEOLOGIC SETTING

Todos Santos strata crop out along a nearly continuous belt extending from western Guatemala (Clemons et al., 1974), along the northern margin of the Chiapas Massif, westward into the western margin of the Veracruz basin, and into the Zongolica foldbelt (Fig. 2). The nonmarine Todos Santos Formation was deposited in extensional basins during the early stages of continental rifting in the Gulf region. Elsewhere in Mexico, comparable sequences of nonmarine detrital strata have been assigned to La Boca and La Joya formations of the Huizachal Group in the Tamaulipas area (e.g., Mixon et al., 1959; Michalzik, 1991), or the Cahuasas Formation in the Huayacocotla area (Ochoa-Camarillo et al., 1998). Todos Santos strata were directly deposited on the basement of the Maya Block, as represented by the Permian metamorphic and plutonic complex of the Chiapas Massif and Paleozoic rocks of the Altos Cuchumatanes. The Todos Santos Formation in Chiapas is overlain by the Upper Jurassic–Lower Cretaceous San Ricardo Formation (Meneses-Rocha, 2001). Here, Todos Santos consists of polymictic conglomerate, sandstone, mudstone, volcaniclastic deposits, and volcanic rocks. The sequence was deposited in alluvial fan, fluvial, and lacustrine environments (Blair, 1987). In the study area of the Sierra Homocline province of west-central Chiapas, Godínez-Urban et al. (2011) report a Pliensbachian U-Pb age for volcanic rocks below the Todos Santos sequence (191 ± 3.0 Ma) assigned to La Silla Formation, and detrital zircons in the range from 196 to 161 Ma in the lower member of the Todos Santos Formation. La Silla Formation is correlated with the Pueblo Viejo andesite (Meneses-Rocha, 1985); they have similar stratigraphic positions and isotopic age determinations. Godínez-Urban et al. (2011) proposed subdividing the Todos Santos Formation into a lower volcaniclastic unit named El Diamante Member and an overlying arkosic member named Jericó.

The term Sierra Homocline for the area east of the Chiapas Massif (Fig. 2) is somewhat misleading, as the Todos Santos Formation around the village of Independencia (Fig. 2A) is exposed within NW trending folds, perhaps associated with the Chicomuselo uplift—an oblique extension of the Chiapas Massif southeast of the study area. Also, two major strike-slip faults are recognized in this sector of the Sierra Homocline. One, the trace of which is north of La Angostura Lake, trends WNW; the other is a conjugate structure that trends north-south, and crosses the study area near El Diamante (Fig. 2A). The homoclinal structure characterizes outcrops of Cretaceous carbonate rocks at La Angostura Lake, and east of it deformation of Neogene age formed the Chiapas foldbelt (Meneses-Rocha, 2001).

The Todos Santos Formation in the Tehuantepec Isthmus, in the Mixtequita region, overlies basement rocks of the Grenville age Guichicovi Complex (Weber and Kohler, 1999; Fig. 2B). It consists of polymictic conglomerate, sandstone, and mudstone. Volcanic rocks have been recognized in the Mesozoic sequence east of the isthmus in the Uzpanapa river area (Herrera-Soto and Estavillo-González, 1991). South of Matías Romero, the Todos Santos Formation is in thrust contact with an allochthonous Upper Cretaceous sequence of metasedimentary and metavolcanic rocks (Fig. 1B; Pérez-Gutiérrez et al., 2009). These authors report detrital zircon data from sandstones in the Todos Santos sequence indicating a Late Triassic maximum deposition age. In the isthmus, Todos Santos is overlain by the Upper Jurassic Mogoñé Formation (Herrera-Soto and Estavillo-González, 1991).

METHODS AND SAMPLING

In the La Angostura Lake area of the Sierra Homocline east of the Chiapas Massif, we collected a total of 34 paleomagnetic sites in rocks of the Todos Santos Formation and the underlying La Silla volcanic sequence. Most of the sites were collected from the volcanic-bearing basal strata assigned to La Silla Formation and from a sequence of intercalated red sandstone and mudstone of the lower Todos Santos El Diamante Member (Fig. 2A). Four sites were collected from Concordia facies of the Jericó Member (high in the Todos Santos Formation), and three sites were collected in conglomeratic sandstones of the Jericó Member of the Todos Santos Formation. In addition, we sampled 11 undeformed dikes of intermediate composition. These intrude rocks of the Chiapas Massif and yield Early to Late Jurassic 40Ar-39Ar dates (Godínez-Urban et al., 2011). Most of the dikes were collected along the road from El Diamante to Custepec (Fig. 2C), additional sites were collected in the valley of the Tablón River, but all but one failed to provide useful results. Finally, five sites were collected in the Tehuantepec isthmus, near localities reported by Molina-Garza et al. (1992) along the Trans-Isthmus highway outside of Matías Romero (Fig. 2B), where Todos Santos strata are gently tilted to the north.

Most samples were obtained with a gas-powered drill, but a few site collections consisted of oriented hand samples, from which standard cores 2.5 cm in diameter were prepared in the laboratory. Samples were oriented with a magnetic compass and clinometer, and where possible with a sun compass. Sites were selected from localities where bedding attitudes could be clearly discerned, but for a handful of sites in andesite flows of the La Silla Formation bedding attitudes are difficult to determine. In those cases, we combined information from nearby sites with good bedding control with information from anisotropy of magnetic susceptibility. In most cases, we observed that the bedding inferred from nearby sites in sedimentary rocks coincided with the orientation of the magnetic susceptibility foliation plane.

For measurements of the natural remanent magnetization (NRM) we used a 2G Enterprises DC-SQUID–based superconducting magnetometer or a JR5 spinner magnetometer. Both instruments are hosted in shielded rooms, at the University of New Mexico and the Centro de Geociencias, respectively. The samples were subjected to progressive alternating field (AF) and thermal demagnetization. The vector composition of the NRM was determined from visual inspection of orthogonal demagnetization diagrams (Zijderveld, 1967), and the directions of components present were determined using standard principal component analysis (Kirschvink, 1980). Site means and formation means were calculated assuming the distribution of directions is reasonably described by Fisher statistics.

PALEOMAGNETIC RESULTS

The magnetization in rocks of the Todos Santos Formation is relatively complex, as the NRM is generally multivectorial. Typically, stable endpoint behavior was observed in volcanic rocks (Fig. 3A) and some of the dikes. Volcanic rocks generally display nearly univectorial behavior, and the magnetization unblocks over a narrow range of laboratory unblocking temperatures typically above 650 °C; demagnetization reveals a northwest-directed magnetization of negative inclination. Other volcanic rocks from the Angostura region have two-component magnetizations, one component is north directed and of low coercivity, the second one is west directed and of negative inclination (in situ). The high magnetic stability component (high laboratory unblocking temperature) shows coercivities in excess of 120 mT, but can be well determined by stable endpoints (Fig. 3C).

In sedimentary rocks interbedded with volcanic flows, a low stability magnetization is north directed and of moderate positive inclination. This component is removed after heating to ∼400 °C (Fig. 3B). The high stability magnetization is of similar direction to that observed in volcanic rocks, but of lower unblocking temperatures (<600 °C). The antipode of the northwest-directed magnetization of negative inclination is observed in other samples, after removal of a north-directed overprint (Fig. 3D). A more complex behavior is observed in some of the red bed samples. For instance, samples from site vc6 are characterized by a low stability, north-directed, and moderate positive inclination (Fig. 3E). An intermediate laboratory unblocking temperature magnetization is southeast directed and of negative inclination. The highest magnetic stability magnetization is southeast directed and of moderately steep positive inclination. Demagnetization data show that the intermediate and high unblocking temperature magnetizations can be removed simultaneously (Fig. 3F, from the same site vc6). The composite magnetization is of shallow inclination; it is an artifact of removing two magnetizations with opposite inclinations.

Near univectorial decay of the NRM is observed in sample vf52 (from the Matías Romero region, Fig. 3G) and in most samples from this region. The magnetization is of distributed laboratory unblocking temperatures between ∼200 °C and 660 °C, and it decays abruptly between 660 °C and 670 °C. The Custepec dikes (Fig. 3H) have high NRM intensities (typically ∼0.2–1.3 A/m), but their behavior is multicomponent. A spurious low-stability magnetization is removed after heating to 200 °C; further treatment removes a prominent north-directed and positive inclination magnetization finally revealing a small, but well-defined, southeast-directed magnetization of negative inclination. We note that the host rock to the dikes yields characteristic magnetizations (ChRM) that resemble previous results for the Chiapas Massif (Molina-Garza et al., 1992), suggesting that regional remagnetization of the dikes and host rocks has not occurred.

The paleomagnetic results for three sites in the red beds of the Todos Santos Formation were lost, and three additional sites did not respond to demagnetization producing uninterpretable data. A ChRM was obtained for all remaining sites (Table 1). Of the sites in dikes along the Tablón River, four did not yield useful data. As mentioned above, the few sites collected in the coarse sandstones of the Concordia facies of the Jerico Member, high in the Todos Santos succession, failed to produce interpretable data.

Site vc18 in La Silla Formation is of particular interest, because samples were collected from discrete volcanic clasts in a volcaniclastic deposit. The clasts are angular, and supported in a sandy matrix of volcaniclastic material; the deposit is interpreted as a lahar. The samples yield univectorial demagnetization behavior, with magnetizations of distributed coercivity and distributed laboratory unblocking temperatures. The direction of the remanence is widely scattered among clasts, but well grouped within single clasts (Fig. 4). This result indicates that acquisition of the magnetization preceded incorporation in the deposit. This is interpreted as a positive intraformational conglomerate test.

DISCUSSION OF PALEOMAGNETIC RESULTS

Of the 24 sites in the Todos Santos and La Silla Formations for which we report an interpretable site mean direction, most have acceptable statistics. Precision parameter values range between 12 and 533, and most of the sites show low within site dispersion. Nonetheless, four sites were excluded from the overall mean because we are uncertain of their structural attitude. Sites vc1 and vc2, in the northern part of the area, are in isolated exposures of andesite flows. AMS data did not define well-developed foliation planes, thus we assumed that these flows have an attitude similar to that of the Cretaceous rocks of the San Ricardo Formation exposed to the east. This correction, however, fails to bring these sites into agreement with others to the south and it is likely that our assumption is incorrect. A similar situation occurred with sites vc10 and vc11, where we had assumed they have an orientation similar to the nearest site. Sites vc22 and vc29 have reasonably well-determined attitudes, but the data from them are interpreted as outliers possibly recording anomalous field behavior (transitional fields or excursions). All but one of the sites in the Jericó member did not yield interpretable data.

The remaining 17 estimated site mean directions define, in situ, two principal groups of directions. One group falls in the northwest quadrant with negative shallow to steep inclinations, and the other in the southeast quadrant with positive inclinations. It is evident that this magnetization was acquired prior to folding of the sequence because the in situ directions are too steep for Jurassic or younger time (Fig. 5). A prefolding age of magnetization acquisition is supported, but not confirmed, by a small increase in the precision parameter of the distribution of site means from 8.1 to 12 (not statistically significant), and also by tilt-corrected shallow inclinations (∼5°) consistent with the inferred near equatorial setting of the Gulf in Early to Middle Jurassic time. When the data set for the Todos Santos Formation in the Matías Romero region is included in a regional tilt test, the increase in the precision parameter from 4.8 (in situ) to 10 (tilt corrected) provides a positive regional tilt test. In addition, a primary origin of the remanence is supported by a positive conglomerate test (site vc18 in lahar deposits), and by the presence of dual polarity magnetizations. The best estimate of the age of the magnetization characteristic of these rocks is provided by the Pliensbachian date of andesite flows at site vc4 (ca. 191 Ma), and the range of dates in detrital zircons from the Todos Santos assemblage (ca. 196–161 Ma; Godínez-Urban et al., 2011). The volcanic rocks are unquestionably of Early Jurassic age. The sites in detrital sedimentary rocks have maximum ages of deposition of ca. 161 Ma, but this age estimate is only based on a single zircon grain; a more conservative maximum deposition age is provided by a cluster of six concordant zircon dates around 171 Ma.

The accepted site collection from La Angostura Lake includes six sites of reverse polarity and eleven of normal polarity. Normal and reverse polarity magnetizations are ∼19° from antipodality, but yield a positive (rank C) reversal test (McFadden and McElhinny, 1990). The data set for the collection of sites from La Angostura exhibits relatively high between-site dispersion (for the distribution of site mean directions, the angular standard deviation is 23.4°). The dispersion may be attributed to the coarse scale of the structural correction, to uncertainties in the structural correction (including the presence of primary dips in volcanic rocks), and also to tectonism (rapid plate motion or rotation contemporaneous with deposition and magnetization acquisition). We note that there is more scatter in declination than in inclination (the average of inclinations is I = 4.3° ± 6.2, whereas declinations vary from 285.1° to 346.2°). This observation suggests the possibility of apparent tectonic rotation due to plunging fold axes, but structural data do not indicate appreciable plunge to any of the folds investigated. These complications contribute to an increase in the scatter of site means, but probably not in a systematic way. We calculate a grand, structurally corrected mean for the Todos Santos and La Silla formations in the Angostura area of declination (Dec) = 325.0°, inclination (Inc) = 4.6° (Table 1; Fig. 5), which we interpret to reflect (with small uncertainty; α95 = 10.8°) the average Early to Middle Jurassic field direction for the southwestern region of the Maya Block (i.e., the Chiapas Massif).

The data set for the Custepec dikes is small. The overall mean is Dec = 332.0°, Inc = 20.2° (n = 6 sites, Table 1). Because the dikes intrude granitic rocks of the Chiapas Massif, there is no accurate reference to the paleohorizontal and we consider this result a less reliable estimate of the Jurassic field direction for the region. We note, however, that the dikes at Custepec and along the Tablón River strike mostly SE (or NE) and dip steeply to the SW (or NW); dips are ∼75° (Fig. 6). The deviation from vertical is consistent with tilt to the NNE, in a direction similar to that for nearby Cretaceous limestones of the Sierra Homocline. Using a cylindrical best-fit approach to all dike orientations, we calculated the correction that brings the dikes closest to the vertical. The tilt correction of 108/0 (trend/plunge) with 22° of tilt on the dike data brings the observed mean paleomagnetic direction to Dec = 335° and Inc = 5° (Table 1), which is closer to the observed mean of the Todos Santos red beds and El Diamante volcanic rocks. Finally, the small data set for the Todos Santos Formation in Matías Romero, combined with previously published data (Molina-Garza et al., 1992), provides the means to test for the regional integrity of paleomagnetic data. The mean for the Todos Santos Formation in the Tehuantepec region is Dec = 312.9°, Inc = 3.2° (n = 7 sites, Table 1). The mean inclination in the Tehuantepec region is indistinguishable from the result obtained in the La Angostura area, and declinations differ by only 12.1°. The data sets from La Angostura, Custepec, and Matías Romero yield a mean paleolatitude of 2.1°N ± 3.4°.

Corrected locality mean directions were used to calculate paleopoles and their associated errors (dp, dm), assuming a central sampling location at 16.1°N–92.9°W (Angostura area), 16.55°N–93.65°W (Matías Romero), and 15.8°N–92.95°W (Custepec dikes). Because the Maya Block is assumed to have shared plate boundaries with North America, we compare in Figure 7 the observed paleomagnetic poles with the North America reference apparent polar wander path (APWP), after Besse and Courtillot (2002). This comparison allows us to estimate relative motion between these crustal blocks. Actual calculations of rotation (R) and latitudinal displacement (F)—and corresponding errors ΔF and ΔR—are based on Butler (1992).

An important caveat in our analysis is that any comparison with the Jurassic APWP for North America is complicated by continued controversy over the reliability of Jurassic results for the North American craton, in particular during the Middle Jurassic, as well as controversy concerning the magnitude of rotation of the Colorado plateau, where many “reference” paleomagnetic data have been obtained. To further complicate this analysis, the Jurassic is a time when there is rapid motion of the North American (NA) plate and a fast rate of apparent polar wander. The first issue arises from the contrasting paleomagnetic results from northeast and southwest North America (e.g., Van Fossen and Kent, 1990; Bazard and Butler, 1991; Steiner, 2003). Middle Jurassic poles from the northeast part of the continent lie at high latitudes, near the present-day North Pole, which in turn predicts higher paleolatitudes for the NA plate. In contrast, Early to Middle Jurassic poles from the American southwest fall at latitudes near 60°N, and in turn predict lower paleolatitudes. The effect of the reference poles from both regions is smaller in the relative orientation of the plate with respect to the paleomeridian (and thus expected declinations and rotation estimates). The second point relates to contrasting estimates of clockwise rotation of the Colorado Plateau with respect to the craton (Molina-Garza et al., 1998; Kent and Witte, 1993; Steiner, 2003; Kent and Olsen, 2008). These estimates range from ∼4° to over 15°. The plateau has the most complete stratigraphic record of the early Mesozoic in North America, therefore a large number of reference poles have been derived from rocks from this region. Taking these facts into consideration, we calculated estimates of R and F based on the Pliensbachian to Callovian age of the magnetization in the Todos Santos and La Silla formations. We make direct comparison with poles for the Pliensbachian Kayenta Formation (Bazard and Butler, 1991), late Callovian Summerville Formation (Steiner, 2003), the ca. 169 Ma Moat Volcanics from the White Mountains (Van Fossen and Kent, 1990), and the synthetic pole for North America for 170 Ma of Besse and Courtillot (2002). The Kayenta paleopole was restored assuming 7.4° rotation of the Colorado plateau (Molina-Garza et al., 1998). It should also be noted here that Wawrzyniec et al. (2007) modeled the rotation of the plateau in an evaluation of a hypothesis that requires excessively large right-lateral strike-slip displacement on the eastern margin of the plateau during the Late Cretaceous to early Tertiary Laramide Orogeny. Such offset, although unreasonably large (Woodward, 2000) was assumed to balance the concurrent shortening along the northern margin of the plateau in association with development of additional structures responsible for shortening during the Late Cretaceous to early Tertiary. It was found that strain estimates north and east of the plateau are consistent with some 5°–8° of plateau rotation about a proximal Euler pole, or possibly a much smaller rotation about a more distal Euler pole. Again, although a rotation appears to have affected the Colorado Plateau, the magnitude of rotation is modest and generally difficult to document at a high level of confidence with paleomagnetic methods.

The estimates of rotation for the Jurassic rocks we studied are summarized in Table 2. For the Angostura area, they range from ∼7° to 38° (counterclockwise); only the latest Middle Jurassic pole for the Summerville provides a small estimate, but most of the sites in the Maya Block are in rocks older than middle Callovian. The Matías Romero result indicates, on average, ∼12° greater magnitude of rotation than the Angostura area result (range from 19° to 50°, but with greater uncertainty because the data set is smaller; Table 2). The observation of appreciable counterclockwise rotation of Todos Santos Formation strata in the Tehuantepec region is most important because it implies that the rotation observed at Angostura is not a local effect of upper Mesozoic or Cenozoic tectonism, but reflects broader scale regional motion of the western Maya Block (Chiapas and the Tehuantepec region). The ∼35° counterclockwise rotation estimate is smaller than that derived from paleomagnetic data for the Chiapas Massif (Molina-Garza et al., 1992), suggesting that either some motion occurred between the Maya Block and North America prior to opening of the Gulf of Mexico (possibly during a Triassic rift phase), the Chiapas Massif moved independently of the Maya Block prior to Early Jurassic time, or a combination of these processes took place. We explore these interpretations below.

Smaller estimates of post–Early Jurassic rotation of the Chiapas Massif area are obtained based on the mean direction of magnetization obtained from the Custepec dikes (range from 19.4° to 25.7°; a smaller value is obtained comparing with the Summerville Formation). Several explanations are possible. One is that the dikes and thus their characteristic magnetizations are younger than Callovian; for several of the dikes only minimum ages can be determined (Godínez-Urban et al., 2011). A second explanation is that the area sampled along the Custepec road was affected by a more complex tilt than that inferred from our analysis of dike orientations. A third explanation is that the data for the dikes are insufficient to average paleosecular variation.

Four sites established in host rock sampled in the same general locality as the Custepec dikes, and the results from a single site reported by Molina-Garza et al. (2009), provide interpretable data that allow us to revise the mean Permian age paleomagnetic pole for the Chiapas Massif (Table 3). The revised pole is located at 12.3°S–178.6°E, with an α95 of 9.2°. Molina-Garza et al. (1992) argued that paleomagnetic directions from the Chiapas Massif should be corrected for the regional tilt of Todos Santos and San Ricardo strata in the Sierra Homocline province. Their argument is based on the fact that Mesozoic strata directly overlying the granitic rocks of the Massif are tilted by some 12° to the northeast. This correction is supported by our analysis of dike orientations in the Custepec and Tablón river areas, but for consistency we used the regional tilt of Cretaceous strata for the Homocline. Correction for this regional tilt brings the pole to 10.6°S–184.2°E. The revised pole is not significantly different from the paleopole previously published, but the larger data set (n = 16 sites) improves the precision of the pole determination. Our field observations in the Angostura area suggest that internal deformation of the Massif may be more complex than a modest regional tilt to the northeast. This simple relationship was observed in the Cintalapa region, but in the Angostura region we observed depositional contacts between crystalline rocks of the Massif and Jurassic stratified rocks, a normal-fault contact between rocks of the Massif and the Jericó Member of the Todos Santos Formation, a strike-slip fault contact between rocks of the Massif and folded strata of the Todos Santos Formation, as well as an outlier of the Massif along the strike-slip fault parallel to the Angostura Lake.

We are cautious about the level of certainty in our interpretation of the paleopole for the Chiapas Massif, but the paleomagnetic result from these rocks is still important in several ways. If regional tilt has affected the Massif, the magnitude is relatively small and it has a minor effect on the pole position. If internal deformation has affected the Massif this could explain the relatively high dispersion of site mean directions (k = 17.9). Such deformation is unlikely to bias the pole position in a systematic way because the paleomagnetic sites are widespread over nearly the entire length of the Massif. Figure 7 shows the Massif pole positions with and without tilt correction.

An additional consideration in interpreting the paleomagnetic data for the Chiapas Massif is the possibility of a component of rotation that may have occurred in late Cenozoic time. Deformation associated with two structural elements adjacent to the Massif may have contributed to a small Cenozoic rotation. These are the Polochic-Motagua sinistral strike-slip fault system to the south and the Chiapas foldbelt to the east. In particular, shortening within the foldbelt may be attributed to a small counterclockwise rotation of the Massif about a local vertical axis that produced northeast-directed contraction. This rotation may account for part of the difference in the magnitude of rotation between the Massif and the Jurassic rocks studied, but only if this rotation was accommodated by faults that did not affect the Sierra Homocline. Additional paleomagnetic data for younger rocks in the region are necessary to evaluate this hypothesis.

Latitudinal displacement between the Maya Block and North America is more difficult to evaluate, because unlike declination anomalies, which are similar for the different reference poles used, inclination anomalies vary significantly using different cratonic reference poles. Nonetheless, it is useful to compare the observed paleolatitudes for Jurassic rocks of the Maya Block with those observed in northeast Mexico. The available data for the Jurassic La Boca Formation of the Huizachal Group in the Sierra Madre Oriental (Gose et al., 1982) indicate significant vertical-axis rotation between localities along the Huizachal-Peregrina uplift. Because the declinations observed at some localities are nearly east-west and inclinations are shallow, polarity interpretations are ambiguous. Using different polarity interpretations we calculated paleolatitudes between 2.3°N and 6.2°N, which are in excellent agreement with paleolatitudes observed in western Chiapas.

Steiner (2005) reported a paleopole for Silurian intrusions in the Maya Mountains that were affected by a major thermal event in the Early Triassic. The paleopole, located at 16.8°S–186.2°E, is statistically indistinguishable from the Chiapas Massif pole. We interpret this observation to suggest that the Massif and the Maya Mountains were already part of the Maya Block and that some motion of the entire Maya Block with respect to North America took place during the protracted rifting phase, prior to the Late Jurassic drift phase and formation of ocean floor in the deep Gulf. This motion has been suggested by other authors (Mickus et al., 2009), and would account for the inferred differential rotation with respect to North America for the Chiapas Massif (∼70°) and for the Todos Santos-La Silla formations (∼35°).

The possibility of post–latest Paleozoic–Triassic relative motion between the Chiapas Massif and the rest of the Maya Block is not fully resolved. Above we noted that the concordance of paleomagnetic pole positions for the Maya Mountains and the Chiapas Massif could be interpreted to suggest minimal displacement between these crystalline terranes that today are included as parts of the Maya Block. However, we identify two lines of evidence that support appreciable relative displacement between the Chiapas Massif and the rest of the Maya Block in pre-Jurassic time. One is the possibility of a tighter fit of the Maya Block against the northern Gulf margin if the Massif was located elsewhere. As mentioned earlier, other authors have proposed a different location for the Massif. The other considers the relationship between the Massif and the East Mexico arc (Dickinson and Lawton, 2001). Restoration of the Maya Block, into the northern Gulf region, results in the position of the arc rocks of the Chiapas Massif far removed from an associated trench. As Figure 1 shows, this is evident in models proposed by Ross and Scotese (1988) and Pindell and Kennan (2009) but only if the Massif is attached to the Maya Block (something not assumed in either of those reconstructions). The relationship of the Massif having an east-dipping subduction system under western Pangea favors a trench configuration similar to that implied in the model of Dickinson and Lawton (2001); that is, a trench closer to the eastern Mexico Tampico Block—even if other details of the reconstruction by Dickinson and Lawton (2001), including the position of southern Mexico, remain unsolved.

Restoration of some 35° of counterclockwise rotation of the Maya Block—a result derived from averaging rotations estimated from the Custepec, La Silla, and Todos Santos mean directions—about an Euler pole on the eastern Gulf (24°N–81.5°E; Hall and Najmuddin, 1994) brings into alignment volcanic rocks of the La Silla Formation with the WNW trend of the Nazas arc of northern Mexico (Godínez-Urban et al., 2011). A similar rotation brings the La Silla-Todos Santos paleopole for the Angostura region into coincidence with the Early-Middle Jurassic segment of the North American APWP (Fig. 7). The currently available paleomagnetic data suggest that ∼35° of counterclockwise rotation is an accurate estimate of the rotational component of motion of the Maya Block during the Late Jurassic drift phase of the Gulf opening.

Greater precision in reconstructing the Maya Block is difficult to afford with the available paleomagnetic data from the Chiapas area, as well as the uncertainty in the North America reference poles. Our database provided by the most robust results includes mean directions from 3 sites in volcanic rocks, and 14 in the overlying red beds assigned to the Todos Santos Formation; one of those sites is in the Concordia facies (uppermost Todos Santos; Godinez-Urban et al., 2011). Given that the data set is of insufficient quality to separate results from volcanic and sedimentary rocks, we attempted a parametric simulation. The inclinations of volcanic rocks and sedimentary rocks are indistinguishable, but declinations of the sedimentary rocks (∼322°) indicate a slightly greater rotation than the simulated declination from volcanic rocks (∼330°).

Based on maximum deposition ages for the red beds, we interpret this as an indication that most of the rotation occurred after ca. 171 Ma, or 161 Ma if a single zircon age determination is used as an age estimate (Godínez-Urban et al., 2011). This interpretation is consistent with stratigraphic constraints (Maya Block rotation postdates Callovian salt deposition; Pindell and Dewey, 1982; Pindell, 1985; Pindell et al., 2006; Salvador, 1987). Based on the available data, we cannot assess if some Todos Santos strata, the upper Jericó Member for example, were deposited when rotation was already under way. An end-Callovian age of 161 Ma is used in Walker and Geissman (2009).

Rotation, however, was apparently completed by the time of deposition of the lower San Ricardo Formation (Guerrero-García et al., 1990). This unit has been assigned a Late Jurassic–Early Cretaceous age. However, Guerrero-García et al. (1990) centered their paleomagnetic sampling on the lower carbonate or Palo Grande Member of the San Ricardo Formation (Quezada-Muñetón, 1983). According to Alencaster (1977), San Ricardo strata contain Kimmeridgian-Portlandian molluscs and brachiopods at a locality 20 km east of Cintalapa. According to Quezada-Muñetón (1983), the Palo Grande Member may extend into the Oxfordian. Buitrón (1978) reports echinoderms known from the Kimmeridgian in Europe. Finally, Michaud (1988) reports Kimmeridgian dasycladales from limestones of the San Ricardo Formation. Therefore paleomagnetic and biostratigraphic data suggest that rotation ended by ca. 151 Ma (end Kimmeridgian) and thus the episode of rotation is estimated to be ca. 10 Ma in duration. If the Massif and its Jurassic cover are restored to offshore Tamaulipas, displacing volcanic rocks of La Silla Formation ∼800 km, the rate of slip along the Tamaulipas-Golden Lane-Chiapas transform is ∼8 cm/yr. This value is somewhat rapid, suggesting that rotation may have initiated slightly earlier or finished slightly later.

The consequence of employing different Euler poles for rotation of the Maya Block is explored in Figures 7 and 8. Rotation about an Euler pole describes the relative motion between two areas on a sphere—North America and the Maya Block in this case. The motion of the Maya Block with respect to North America is equally expressed by the motion of the paleomagnetic poles for the Maya Block with respect to the North American apparent polar wander path, and this motion describes a small circle on the sphere. The Euler pole proposed by Hall and Najmuddin (1994), for example, restores the Todos Santos-La Silla pole along a circle in the direction of the low-latitude Jurassic segment of the APWP, toward the Kayenta cratonic reference pole. The Euler pole suggested by Ross and Scotese (1988) rotates the Todos Santos-La Silla pole into a position intermediate between the low-latitude and high-latitude reference pole alternatives. We note that the paleomagnetic data from the Chiapas Massif area, Matías Romero area, and Todos Santos-La Silla formations approximate a small-circle, the best fit of which defines an Euler pole at 19°N–265°E. This Euler pole, however, does not provide a satisfactory fit for the Maya Block in the interior of the Gulf region because it places the Maya Block too far to the south in the Gulf region, in an area that was occupied by South America. The confidence error in the determination of the best-fit small circle, however, includes the poles proposed by Hall and Najmuddin (1994) and Pindell and Kennan (2009).

An alternative way of considering the evolution of the Gulf region is by keeping North America fixed (Fig. 8). We place a trench with a configuration similar to that proposed by Dickinson and Lawton (2001), which closely parallels the trend of the Nazas arc of northern Mexico, and then turns abruptly south. We place the northern margin of South America according to Pindell and Kennan (2009). We then reconstruct the Maya Block according to different Euler poles. Although a definitive and robust single Euler pole is not attainable, it appears that Euler poles in the eastern Gulf—near western Cuba—best fit the requirement of reconstructing the Massif at a reasonable distance from the trench, and the apparent polar wander path defined by data for the Maya Block. This path requires an Euler pole close to the Maya Block.

CONCLUSIONS

Volcanic rocks of the Lower Jurassic La Silla Formation and detrital sedimentary rocks of the overlying Todos Santos Formation in west-central Chiapas were deposited on Permian crystalline rocks of the Chiapas Massif. The ChRM in these rocks is northwest directed and of shallow inclination (∼5°). Dual polarity magnetizations, a conglomerate test, and a regional tilt test all suggest that the characteristic magnetization of these rocks is primary. Both volcanic rocks and red beds have similar shallow inclinations. The ChRM of Todos Santos strata in the Tehuantepec Isthmus and mid-Jurassic dikes that intrude crystalline rocks of the Chiapas Massif are also northwest directed and shallow. The paleomagnetic results from these rocks are discordant with respect to the North America APWP, and indicate that the Maya Block has rotated in a counterclockwise sense; our best estimate is ∼35° since ca. 161 Ma, and it is likely that most of this rotation occurred over a time span of ca. 10 Ma. Some of the uncertainty in the estimate of the magnitude of rotation is due to the poor definition of the Early to Middle Jurassic segment of the North America APWP. Regardless, the timing and magnitude of rotation are both consistent with the proposed post-Callovian opening of the Gulf of Mexico, and a rotational origin for the Gulf. Rocks of the Chiapas Massif and the Maya Mountains record a greater magnitude of counterclockwise rotation (∼70°), which in turn suggests that prior to the drift phase of the Gulf of Mexico opening, relative motion between the Maya Block and North America (or between Chiapas and North America) occurred during a protracted rifting phase in the Triassic.

This research was supported by a grant of the Petroleum Research Fund of the American Chemical Society to Wawrzyniec and also by grant IN121002 to Molina-Garza from the PAPIIT-UNAM program. We also thank the assistance in the field by Linda Donohoo-Hurley, as well as the Associate Editor and reviewers, whose comments improved the manuscript.