Synfolding remagnetization and deformation: results from Palaeozoic sedimentary rocks in West Virginia

Summary

Palaeomagnetic analysis was conducted at 27 sites from two stratigraphically adjacent Palaeozoic units along two transects (Meldey and Rig) of the Patterson Creek anticline in the Appalachian orogen of West Virginia to better understand the relationship between deformation and remagnetization. While no primary directions were observed, we isolated identical, well-defined, secondary Late Palaeozoic magnetizations in carbonates from the Late Silurian Tonoloway Formation and the Early Devonian Helderberg Group. Standard application of conventional palaeomagnetic fold tests suggests that the magnetizations are synfolding in the Helderberg Group and pre-folding to early synfolding in the Tonoloway Formation. A fold test on all data combined indicates a synfolding magnetization (best grouping at 60–65 per cent unfolding), but the resulting distributions are not circular. An alternative interpretation, based on optimal differential unfolding at the site mean level yields one group with a nearly pre-folding, or early synfolding, magnetization (>70 per cent unfolding) that is both circular and concordant with North America's apparent polar wander path. The second group has a ‘variable synfolding’ magnetization that is elliptical at the optimum unfolding level and falls off the path. Differences in either the age of remagnetization or viscous partial thermoremanent magnetization contamination cannot account for this result. The sites falling into the pre-folding group are biased towards the Tonoloway Formation, the Medley Transect, the northwest limb, away from small folds and lower dip angle relative to the sites in the synfolding group. The two groups defined by the fold test results also have a relationship with both total strain and strain partitioning. The rocks containing the early synfolding remanence exhibit less total strain and, in particular, less pressure solution strain relative to grain boundary sliding and calcite twinning. It is likely that the remagnetization in the entire area was coincident and occurred prior to folding even though some sites reveal synfolding directions. The variation in the behaviour of the remagnetization relative to folding is unclear at present but it indicates that pressure solution and stylolite formation are important processes in remagnetization. In any event, the relationship between deformation and magnetization is not simple. Apparently identical magnetizations from adjacent sites may have very different histories and/or responses to deformation. While the exact relationship between remagnetization and deformation remains elusive this study demonstrates that significant variation in behaviour can exist at a very local scale.

deformation, palaeomagnetism, Permian, stress distribution, tectonics

Introduction

The occurrence of widespread secondary magnetizations that cannot be attributed to moderate to high-temperature processes has been well documented (e.g. Elmore & McCabe 1991). All potential remagnetizing mechanisms can be broadly grouped into one of the following three processes for the purposes of interpreting fold test results: (1) growth of new magnetic minerals; (2) unblocking and subsequent reblocking in existing magnetic minerals; and (3) physical rotation of magnetic minerals during deformation. Many remagnetizations are associated with synfolding magnetizations, where the palaeomagnetic data from opposing limbs have the tightest grouping at an intermediate level of unfolding. The growth of new magnetic material and unblocking and subsequent reblocking of existing magnetic domains during folding, or between multiple folding events will produce a true synfolding magnetization as long as all pre-existing magnetizations are destroyed. However, the presence of pre-existing magnetizations or grain rotation could yield ‘apparent synfolding’ magnetizations where the level of optimum unfolding does not necessarily relate to the timing of remagnetization.

Several mechanisms have been proposed to account for synfolding results (e.g. Hudson et al. 1989). Possible mechanisms to produce true synfolding magnetizations include orogenic fluids (e.g. Oliver 1986), fluids derived from topographic recharge (Garven 1995), organic matter maturation (Banerjee et al. 1997), diagenetic recrystallization and clay alteration (Katz et al. 2000). Spontaneous thermoviscous remagnetization (e.g. Kent 1985) has been proposed as a mechanism for generating low-temperature remagnetizations that can produce apparent synfolding magnetizations. They can occur when simultaneous demagnetization of two or more overlapping components of differing age produce the best estimate of the fold test that is not indicative of the age of acquisition of the magnetization. Apparent synfolding magnetizations may also arise from improper unfolding caused by errors in measuring structural orientations or a lack of knowledge of the complete structural history. For example, rocks subjected to multiple deformation events can result in a structure today that is difficult or impossible to return to its completely unfolded orientation owing to the absence of knowledge concerning the intermediate structural orientation.

This paper presents a case study examining the relationship between remagnetization and folding in order to determine the true nature of a synfolding magnetization. The study area is a symmetric macroscale fold from the central Appalachian Valley and Ridge province of West Virginia. In our approach to the problem, we will integrate high-quality palaeomagnetic data and detailed strain partitioning on carbonates to test for a connection between remagnetization, strain and lithology.

Geology

The fold and thrust belt of the central Appalachians consists of a series of blind fold and thrust sequences (Gwinn 1964) over several hundred kilometres of strike length (Fig. 1a) with a consistent trend of approximately north 20° to 40° east. The region was described in detail by Wilson & Shumaker (1992), with the aid of seismic profiles. It contains several doubly plunging anticlines and synclines, some reaching over 150 km in length. The Patterson Creek anticline, the focus of this study, follows the main structural trend for over 40 km (Fig. 1b) and is 3–4 km wide. Five conformable stratigraphic units of Palaeozoic age ranging from the Lower Silurian McKenzie Formation to Lower Devonian Oriskany Sandstone are exposed along at least one of two transects, in the Medley quadrangle and further south in the Rig quadrangle. Thus, with excellent exposure cutting across several stratigraphic horizons, the Patterson Creek anticline is an ideal natural laboratory to examine the physical and temporal relationship between macroscopic folding, microscopic deformation and the acquisition of secondary magnetizations.

Figure 1.

Location of the study area showing: (a) a satellite image displaying the broad regional trend of the Appalachian fold and thrust belt; (b) geological map of the Patterson Creek anticline; (c) a cross-section through A–A' showing the sampling locations of the Medley transect; and (d) a cross-section through B–B' showing the sampling locations of the Rig transect.

This study concentrates on a detailed analysis of the Tonoloway Formation and the conformably overlying Helderberg Group (Head 1974), two carbonate units, exposed along both the Medley and Rig transects of the Patterson Creek anticline. Additional palaeomagnetic results from the McKenzie, Williamsport and Orsikany Formations and a carbonate vein from the same transects can be found in Lewchuk et al. (2002).

The Tonoloway Formation and Helderberg Group were remagnetized (Evans et al. 1999; Lewchuk et al. 2001, 2002; Elmore et al. 2001) during the Permian Kiaman event (Van der Voo 1993, and references therein) and have locally experienced the same deformation history and pressure–temperature–fluid conditions (Evans & Battles 1999). Limestones of the Helderberg Group contain a wide range of lithologies including grainstones, wackestone and carbonate mudstones (Dorobek & Read 1986; Dorobek 1987; Meyer & Dunne 1990), but are dominated by coarse-grained lithologies. In contrast, the Tonoloway Formation is dominated by fine-grained lithologies such as carbonate mudstone, although coarser-grained rocks such as wackestones and occasional packstones and grainstones are also present (Reger 1924). In outcrop, the rocks appear ‘undeformed’ and exhibit common to pervasive bed-parallel stylolites that are spaced on the order of centimetres to tens of centimetres. Bed-normal stylolites are rare to very uncommon. The Tonoloway Formation ranges in thickness from approximately 130 to 180 m and is conformably overlain by the Helderberg Group, ranging in thickness from 80 to 120 m (Dorobek & Read 1986; Dorobek 1987; Meyer & Dunne 1990).

Methods

Palaeomagnetism

For the palaeomagnetic sampling, seven to ten oriented 2.5 cm diameter cores were drilled in the field and oriented in situ with a magnetic compass at 27 sites along two road transects (Medley and Rig) across the Patterson Creek anticline (Fig. 1). Both units are well exposed at multiple locations in the middle of the flanks, away from the hinge zone. The samples from each site were restricted to a single bed of less than 0.5 m thickness and all cores were taken within 1 m of each other along strike.

All specimens were measured on a three-axis 2G cryogenic magnetometer housed in a shielded room. One specimen from each oriented core was subjected to detailed thermal demagnetization in at least 12 steps up to 600 °C using a shielded Schonstedt furnace, also housed in the shielded room. Linear segments of the magnetization in each specimen were visually identified on orthogonal vector component plots (As & Zijderveld 1958) and calculated using a modification of the least-squares fitting technique (Kirschvink 1980). Site means were calculated using the bivariate extension (LeGoff 1990; LeGoff et al. 1992) of the traditional Fisher (1953) statistic.

Progressive tilt/fold tests were conducted on both units at the site mean level, in 2 per cent increments from post-folding (geographic) coordinates with pre-folding (stratigraphic) coordinates, using four statistical methods. To assess the fold test we used the traditional precision parameter (κ) of Fisher (1953), because it is easily comparable with previous studies. We also report the F statistic test of McFadden & Jones (1981) (hereafter M&J) because it provides confidence limits for the results. Unfortunately the M&J confidence estimate is strongly dependent on N (the number of sites), thus it can yield large confidence limits when the test is conducted at the site mean level. Therefore, we also used the parameter estimation method of Watson & Enkin (1993) (hereafter W&E), which uses traditional Fisher statistics accompanied by a bootstrap resampling scheme at the site mean level to provide 95 per cent confidence limits concerning the best estimate of unfolding. It is complimentary to the M&J test because the confidence limits for the W&E test are dependent on κ rather than on N. Finally, we used the bivariate (BV) extension of the Fisher statistic because it provides an estimate of the shape (ellipse with perpendicular axes k_x and k_y, where k_x≥k_y) of a given data set with respect to an ideal Fisher distribution (k_x/k_y= 1.0). The deviation from circular symmetry can then be assessed using a standard F test with 2(N− 2) and 2(N− 2) degrees of freedom (LeGoff et al. 1992). In principle, a set of data that fails this test cannot have been drawn from a single Fisherian distribution. Thus, if a combined population fails the BV test for circular symmetry it can also be considered to have failed the fold test at that level of unfolding.

Strain partitioning

A detailed examination of the microstrain data and its comparison with the anisotropy of magnetic susceptibility and the anisotropy of remanent magnetization can be found in Evans et al. (2002) so only a brief description of the technique and results is presented here. For strain analysis, oriented block samples were collected at each site from within the same layer cored for magnetic samples. Block samples were taken as close as possible to the core holes and usually incorporated the cored rock. For each block sample collected, three mutually perpendicular thin sections were cut and polished to 0.03 μm alumina. These were examined by optical and reflectance petrographic methods in the transmitted plane and polarized light to determine the minerals present and characterize the deformation mechanisms operating in the rock.

Strain partitioning was performed to evaluate the relative magnitude of the deformation mechanism(s) that are responsible for the strain in the rocks and to examine the relationship between rock strain and magnetic response. Strain partitioning (Ramsay & Huber 1983; Groshong et al. 1984; Wu 1989; Evans & Dunne 1991; Couzens et al. 1993; Onasch 1994; Harrison & Onasch 2000) is an effort to separate the finite strain into components caused by different deformation mechanisms. For limestones deformed at low temperatures (<200 °C) as in this study, finite strain may be partitioned into: (1) a compaction pressure solution with pressure solution surfaces parallel to bedding; (2) a tectonic pressure solution with pressure solution surfaces at a high angle to, or normal to bedding; (3) a calcite twinning strain; (4) an intragranular strain resulting from dislocation mechanisms; and (5) an intergranular strain resulting from diffusion accommodated grain boundary sliding (GBS).

The deformation mechanisms that are active during deformation, and that are responsible for accumulating rock strain, are a function of pressure, temperature, strain rate and grain size. Changing any one of these parameters may result in a different mechanism taking over in a rock. Therefore, even in a single fold, where the rocks are all deforming under the same pressure, temperature and strain rate conditions, rocks of differing grain size will acquire strain in different ways and in different magnitudes. Even within a single outcrop, a coarse-grained layer will deform differently from a fine-grained layer.

Partitioning pressure solution strain. There are several techniques that may be used to estimate the volume loss caused by a pressure solution at the thin-section scale. For stylolitic pressure solution strain, stylolite amplitude (Stockdale 1926; Bodou 1976; Silbey & Blatt 1976; Smart et al. 1997) may be used. In this method, the amplitude of the stylolite gives a minimum estimate of volume loss normal to the stylolite surface.

Alternatively, concentrations of relatively insoluble mineral grains (pyrite, quartz or dolomite) may be used to determine volume loss (Heald 1956; Silbey & Blatt 1976). In this study, the distribution of authigenic quartz in the matrix was compared with the concentration of quartz in the stylolite zone. Up to ten scan lines were made across a photomicrograph mosaic and the amount of quartz (in microns per micron of scan length) in the matrix was compared with the amount of quartz in the stylolite zone. The volume loss determined in this manner compared favourably with that determined from stylolite amplitude measurements on the same stylolite. However, this method only gives a minimum value for volume loss (shortening) because quartz is also pressure solved within the stylolite zones. Therefore, where authigenic quartz was present, both methods were used, where authigenic quartz was lacking, only stylolite amplitude was used. Pyrite and iron oxide grains were not used because in some cases they have undergone considerable dissolution and reprecipitation as described above.

In cases of pervasive grain-to-grain pressure solution, the normalized Fry method (Fry 1979; Erslev 1988) was used to determine total finite strain. Any component of twinning strain in the calcite grains was then subtracted and the remaining shortening attributed to pressure solution. Pressure solution may be subdivided into compaction pressure solution (CPS) with the pressure solution zoned parallel to bedding and tectonic pressure solution (TPS) with the pressure zone normal to bedding and parallel to strike.

Partitioning strain caused by twinning, grain boundary sliding, and dislocation glide and climb. Strain caused by calcite twinning (TW) may be determined from the calcite strain gauge technique (Groshong 1974; Evans & Groshong 1994). Grain boundary sliding (GBS) strain can only be determined in fine-grained rocks where there are passive markers such as pellet outlines. The Fry method is used to determine finite strain. If no other mechanisms were operative, the total strain may be attributed to GBS. In calcite rocks deformed at temperatures <200 °C, grains deformed by dislocation mechanisms are uncommon and, consequently, the contribution to the total strain is minor. The presence of grains distorted by dislocation mechanisms is simply noted.

Palaeomagnetic results

The specimens were extremely well behaved during demagnetization and all had remarkably similar behaviours. After removal by approximately 300 °C of a soft, viscous, overprint that is parallel to the Earth's present magnetic field direction, all samples exhibited a sharp bend followed by linear demagnetization to the origin (Fig. 2). Most specimens were completely demagnetized by 500–550 °C and all specimens were completely demagnetized by the Curie temperature of magnetite. Excellent data were obtained from 216 of the 218 specimens measured. Individual specimen maximum angular deviation or MAD angles (Kirschvink 1980) rarely exceeded 10° and most were less than 5°. Site mean directions exhibit very tight within site consistency (Table 1) with κ > 100 for most sites. Initial examination of the data was conducted by grouping along the sample transect, and then by formation.

Figure 2.

Normalized orthogonal demagnetization plots in geographic coordinates of example specimens representative of the study area. Circles are projections in the horizontal plane, while squares are projections in the vertical plane. Treatments are in °C. Note that all of the specimens are virtually identical in demagnetization behaviour in terms of both the unblocking temperatures and the relative sizes of the VRM and ChRM.

Table 1.

Site mean statistics for the lithological units before (geographic) and after (stratigraphic) correction for folding. n/n_o are the number of specimens used versus demagnetized, str and dip are the strike and dip of bedding, dec, inc, k and α₉₅ are the declination, inclination and precision parameters of Fisher (1953), ODU is the optimal differential unfolding level for each site, CPS, TPS and PS refer to the compaction, tectonic and total pressure solution determinations expressed as per cent volume loss, ESF and VSF refer to the folding behaviour (see text for explanation).

Medley transect

The fold axis at Medley plunges 2° towards 206°, based on the strike and dip measurements at each palaeomagnetic site. Five sites were taken in the Helderberg Group and six sites were taken from the Tonoloway Formation.

The two stratigraphic units gave very different fold test results. The Helderberg data yielded a clear synfolding result with maximum estimates ranging from 49 to 56 per cent unfolding using all fourfold tests with confidence limits clearly excluding the possibility of either a pre- or post-folding remanences (Table 2, Figs 3a and 4a). This result is virtually identical to twofold tests on the Helderberg from the Broadtop anticline (Elmore et al. 2001), further east in West Virginia (Fig. 4a). Data from the Tonoloway Formation, on the other hand, yielded nearly pre-folding magnetizations with maximum estimates ranging from 86 to 104 per cent unfolding (Table 2, Figs 3b and 4b). The fold test results from the Tonoloway Formation yield a magnetization with a much higher level of unfolding than any of the results for folds in the Helderberg Group in this region (Fig. 4a).

Table 2.

Fold test results for the various groups of data conducted at the site mean level. Dec. and Inc. are the aggregate estimate for optimum unfolding. The actual values may vary by a fraction of a degree for each method of computation. Kappa values are given for maximum grouping (τ_xx). M&J, W&E and BV are all expressed as best estimate with 2π confidence limits or ‘dnp’ which means it did not pass at any level of unfolding. ESF and VSF refer to early and variable synfolding behaviour.

Figure 3.

Expanded, equatorial stereonet projections of the site mean directions for both lithologies at each transect. Open (shaded) ovals are the 95 per cent confidence limits in geographic (stratigraphic) coordinates defined by the bivariate extension (LeGoff 1990; LeGoff et al.1992) of the Fisher (1953) statistic. (3e) Apparent Polar Wander Path (Van der Voo 1991, 1993) with mean poles (stars) and 95 per cent confidence width from Early Carboniferous (1 = eC) to Late Triassic (6 = l Tr) showing the Helderberg (Tonoloway) poles in black (grey) for the Medley (M) and Rig (R) transects circumscribed by their cones of 95 per cent confidence (Khramov 1987).

Figure 4.

Incremental fold test results for the following locations: (1) and (2) Helderberg at the Romney Quarry and Rocks locations on the Broadtop anticline (Elmore et al. 2001); (3) Helderberg at Medley; (4) Tonoloway at Medley; (5); Helderberg at Rig (6) Tonoloway at Rig and (7) alternate ‘pre-folding’ group.

The mean directions at 50 per cent unfolding for the five Helderberg sites α₉₅= 4.7°) yields a pole position at 125.1°E, 53.6°N (A₉₅= 3.4°, (Khramov 1987) and at 90 per cent unfolding the six Tonoloway sites (α₉₅= 4.3°) yield a pole at 122.5°E, 48.9°N (A₉₅= 3.1°). When these directions are compared with the North American apparent polar wander path (APWP) (Fig. 3e) the ages of the magnetization are similar. The Tonoloway pole yields a perfect fit to the APWP for the Permian, falling between the reference poles for 267–281 and 246–266 Ma (Van der Voo 1991, 1993). The Helderberg mean pole falls slightly off the APWP but it is closest to the 246–266 Ma reference direction. Thus, both units yield Late Palaeozoic reversed magnetizations typical of Palaeozoic sediments from the eastern United States and diagnostic of the Late Palaeozoic Kiaman reversed superchron.

Rig transect

The fold axis at Rig plunges 1° towards 217°, based on the strike and dip measurements at each of the six sites taken in the Helderberg Group and 11 sites from the Tonoloway Formation.

Two minor folds in the Tonoloway Formation were sampled along this transect. Each had complete exposure so we could collect identical beds across the fold. Sites 69–72 come from a fold of approximately 100 m wavelength and sites 34 and 35 come from a fold of approximately 25 m wavelength. Both small folds yield poorly constrained synfolding magnetizations of approximately 50 per cent unfolding (Table 2). Curiously, the two minor folds yielded the lowest estimates for unfolding in this study.

The formation-based fold test results here differed from that of the Medley Transect. At Rig the Helderberg data yielded an optimum result similar to but at a slightly higher level of unfolding (68–71 per cent) than seen at Medley (Fig. 4b). The Tonoloway data gave a lower level for optimum unfolding (59–61 per cent) (Table 2) that is different from the Tonoloway data at Medley (Fig. 4b). At optimum unfolding the Helderberg data (N= 5, α₉₅= 4.7°) combine to give a pole at 118.9°E, 52.7°N (A₉₅= 3.3°) and the Tonoloway (N= 11, α₉₅= 2.2) pole is at 122.3°E, 50.9° (A₉₅= 2.2°). Similar to the poles for the Medley transect, these two poles are concordant with the 246–266 Ma reference direction. The 95 per cent confidence limits for all four pole positions overlap such that their relative ages are indistinguishable based on comparison with the APWP (Fig. 3).

An important characteristic of the Tonoloway data from the Rig transect is that it fails the BV test for circular symmetry at all levels of unfolding for both the two small folds and all sites combined. This provides an indication that the fold test may not be applicable here.

Combined data

Analysis of all data, grouped by either formation or transect yields similar, synfolding results in all of the fold tests (Table 2). The 10 sites in the Helderberg Group have an optimum unfolding result ranging from 60 to 68 per cent, while the 17 sites in the Tonoloway range from 63 to 71 per cent. All 11 sites from the Medley transect combine to yield an optimum unfolding level of 61–62 per cent, while all 16 sites combined at Rig have an optimum unfolding of 62–64 per cent.

With such similar results from the combined fold tests one could draw the simple conclusion that all sites were remagnetized at approximately the same time approximately 60–70 per cent through the deformation process. However, several problems exist with such an interpretation. First, it requires that the disparate results of the fold tests for different sites be rejected as statistical artefacts. Secondly, the distribution of site mean data at optimum unfolding differs markedly from a Fisherian distribution and fails the BV test for all 27 sites at optimum unfolding (63 per cent unfolding kx/ky= 3.1, F_crit= 1.5(2σ)). This shows that application of the fold test has not, in fact, simply removed the post-magnetization deformation of a magnetization acquired during folding. This is demonstrated visually by a plot of the site mean data for all 27 sites at the optimum result of 63 per cent unfolding (Fig. 5). Note that the distribution of the site means is still elongated along the plane of unfolding. If the fold tests had performed as intended then the dispersion in this distribution should have been approximately circular.

Figure 5.

(a) All 27 site means plotted at optimum (63 per cent) unfolding. For reference the white ovals are the overall mean directions for each limb in geographic coordinates. Projection and other symbols are as in Fig. 4. Note that the site means are skewed along the plane containing the initial limb means in geographic coordinates. (b) Site mean optimal differential unfolding levels.

Alternative grouping

The failure of the test for circular symmetry at optimum unfolding argues for an alternative method to the traditional grouping of the data by formation. To better evaluate the data at the site mean level we analysed the site means using the optimum differential untilting (ODU) approach (Enkin et al. 2000) based on the fold test method suggested by Shipunov (1997). In this modification of the fold test each site was assigned an individual unfolding level based on the small circle intersections of all sites with the results for each site restricted to between 0 and 100 per cent unfolding. The results of this test are shown in Fig. 5(b) and Table 1 where two behaviours are observed. The first group has 11 sites with a variable synfolding (VSF) behaviour ranging from 0 to 60 per cent unfolding and the second group has 16 sites with a pre-folding to early synfolding (ESF) behaviour, all having ODU levels of above 70 per cent.

Close examination of the data in stratigraphic coordinates shows the differing behaviours of the two groups. The ESF group does not cross over itself during unfolding and forms a roughly circular group with an inclination close to 0°, and concordant with the APWP (Fig. 6a) similar to the six Tonoloway sites at Medley. The remaining sites (VSF group) cross over during unfolding (Fig. 6b) and yield discordant directions in stratigraphic coordinates.

Figure 6.

Proposed alternative grouping into: (a) early synfolding and (b) variable synfolding behaviour based on the site mean data (grey ovals) in stratigraphic coordinates. The stars are the stable craton apparent polar wander reference directions from 282 to 308 Ma (bottom right) to 233–245 Ma (top left) (Van der Voo 1993).

A fold test conducted on the ESF group yields an optimum magnetization direction (N= 16) of D= 166.0°, I=−0.9° (Table 2, Fig. 4b) that can be considered to be pre-folding according to the M&J and BV tests plus almost pre-folding (85 per cent ±3) using the W&E test. The failure of the W&E test may be caused by the unusually high kappa values at the site mean level (Table 1) and the resultant low within site dispersion. The nature of the W&E test results in very narrow pass windows when kappa is very high.

A fold test on the VSF group is not as clear. It yields an optimum fit that indicates a synfolding magnetization (N= 11) with a mean of D= 168.1°, I=−5.7° at approximately 54 per cent unfolding (Table 2). However, at any level of unfolding, it fails the BV test for circular symmetry (Table 2) and the M&J test is not applicable because one limb has only one site. Thus the fold tests are indicating that this was not originally a Fisherian distribution with a common deformation history.

The question then becomes, what is the explanation for these two separate behaviours? Examination of the relationship to stratigraphy shows some trends but no clear distinctions. The ESF group is dominated by the Tonoloway (11 of 16 ESF sites), while the VSF group is evenly split between the two lithologies. The Medley transect has eight of 11 sites in the ESF group while the Rig transect has eight of 16. 11 of 12 sites with northwest dip directions are in the ESF group, while only five of 15 sites with southeast dips are in the ESF group, although four sites (35, 51, 71 and 72) on minor folds have local dip directions that are opposite to the regional limb orientation. The dip angles differ slightly with the ESF group having an average dip of 32°± 7°, while the VSF group has an average dip of 41°± 14°. Seven sites (32, 34, 35, 69, 70, 71, 72) are known to be near minor folds (≲100 m wavelength) and four of these (34, 35, 71, 72) are in the VSF group. Other sites may have been on minor folds that were not evident owing to lack of exposure. Thus, there is no clear independent geological or geographic justification for such a split of the palaeomagnetic data into two subsets.

Strain partitioning results

Fig. 7 shows a series of histograms summarizing the partitioned strain data for the palaeomagnetic sites. The histograms are grouped according to demagnetization behaviour, into the ESF and VSF groups. In general, several observations may be made. First, the rocks that exhibit VSF behaviour have higher CPS strain (7.8–22.0 per cent shortening) compared with those that have an ESF behaviour (0.0–14.0 per cent shortening) with the exception of site 68. Similarly, TPS strain is higher in VSF samples (up to 13.3 per cent shortening) than in ESF samples (up to 7.7 per cent shortening). Site 68 is the only one to have an ESF magnetization behaviour, yet anomalously high CPS and TPS values characteristic of the VSF group (Fig. 7). Secondly, the rocks with the least amount of strain (samples 18, 36, 39 and 44) are mudstones that are deformed primarily by GBS and have little to no pressure solution or twinning strain. These sites exhibit an ESF behaviour. Thirdly, calcite twin strain magnitudes (0.5–3.4 per cent shortening) are similar throughout the fold and there is no significant correlation between twin strain magnitude and unfolding behaviour. However, the presence of distorted calcite grains indicate that dislocation mechanisms were active in three VSF samples (66, 67, 72), but in only one ESF sample (13). Finally, the rocks with the highest total strains are generally those on the southeast limb of the fold (Fig. 1), while the rocks on the northwest limb have a smaller total strain.

Figure 7.

Results of strain partitioning with the ESF group in the top panel and the VSF group in the bottom panel. CPS, compaction pressure solution; GBS, diffusion accommodated grain boundary sliding; TPS, tectonic pressure solution; TW, calcite twinning strain. DGC, presence of distorted calcite grains indicating dislocation mechanisms were active. SE and NW indicate limb of Patterson Creek anticline sample is from. Dhl is Helderberg Group and Sto is Tonoloway Formation.

In summary, the unfolding behaviour can be correlated to the amount of strain in the rocks when strain partitioning is conducted. In particular, the amount of pressure solution strain, both compaction and tectonic, seem to control the behaviour. There is a lesser contribution by other deformation mechanisms.

Discussion

The disparity in the results of the fold tests presented here is problematic given that the two units contain such similar magnetizations, both in demagnetization behaviour and apparent age. Many researchers have proposed that such synfolding results are the product of something other than a simple single remagnetization event during folding (e.g. Hudson et al. 1989; Borradaile 1997). Three possibilities exist: (1) the remagnetization has been contaminated (i.e. it is the vector addition of two components); (2) it was not acquired simultaneously across the study area; and (3) it did not behave as a passive marker during folding.

Potential for contamination

Kent (1985) and Hudson et al. (1989) have argued that moderately elevated burial temperatures can produce a modern, but laboratory-resistant viscous partial thermoremanent remagnetization (VpTRM) that may be responsible for some of the synfolding results. This behaviour could lead to overlapping coercivity spectra of a pre-folding magnetization and a post-folding magnetization that could appear to be a synfolding magnetization. Borradaile (1999) showed that apparently single-component linear directions isolated in Mesozoic carbonates of eastern England may, in fact, be the vector addition of a modern component with an ancient one. He showed that a spontaneous viscous component acquired during the most recent chron (<0.7 Ma) may take up to 300° of laboratory demagnetization to be removed when it should have been removed by approximately 150° according to the theoretical work of Pullaiah et al. (1975). Middleton & Schmidt (1982) published an alternative set of time–temperature relationships that propose much higher laboratory unblocking temperatures for VpTRMs. However, subsequent work by Williams & Walton (1988) and Worm & Jackson (1988) confirmed the validity of the Pullaiah et al. (1975) curves. Recently Dunlop & Özdemir (1993) showed that the discrepancy between the two sets of type curves may reflect the varying thermal demagnetization responses of single- (SD) and multidomain (MD) magnetites. While true SD magnetite conforms to the theoretical curves of Pullaiah et al. (1975), MD magnetite may have unusually high laboratory unblocking temperatures. Therefore, in field studies, where mixed populations are almost certain to occur, the role of a laboratory-resistant VRM must be considered.

Contamination of the magnetization by a resistant VpTRM or an underlying older magnetization can be tested by examination of the geometrical relationship between the two components and the folds. If a fraction of this component unblocked in the same temperature range as the ChRM during laboratory demagnetization then one could produce linear segments from the two overlapping magnetizations that appear to be a single magnetization, yielding a false synfolding result. However, since both the primary and modern field directions are known then the overlapping components should have a predictable geometry. For example, a resistant modern component combined with a pre-folding component would produce results that are biased towards the modern field direction and the amount of bias would be greater for those locations that give fold tests closer to post-folding. Examination of the geometry of the two groups relative to the PEMF shows that this cannot be the case in this study. The ESF group has an inclination that is horizontal, while the VSF group has a similar declination but one that is slightly upwards. Thus the group that is most likely to be contaminated (VSF) retains directions that are further from the PEMF direction so we cannot explain the differences between the ESF and VSF groups by VpTRM contamination. Contamination by a secondary, post-folding but ancient, chemical remagnetization cannot be ruled out. However, there is no evidence for the existence of such a component in this area, and its effect cannot be evaluated in the same way as the VpTRM without some idea of its remagnetization direction.

If the data were contaminated by an older component (presumably primary, but potentially having any direction) the combination of the two directions should yield an incorrect but still pre-folding result because any component older than the one isolated in the ESF group must also be pre-folding. Again this is not the case with our data. The poles at optimum unfolding shown in Fig. 3 do not exhibit any trends with respect to the level of unfolding.

Finally, the demagnetization characteristics of these specimens do not favour the presence of overlapping components. In all cases an abrupt change of direction at approximately 300° (Fig. 2) and geologically reasonable directions for each component imply that the two components isolated do not have overlapping coercivity spectra. Thus the possibility that contamination by either younger or older components explain the two groupings can be rejected.

Potential for different remagnetization ages

Both units retain magnetizations in magnetite with similar demagnetization behaviours and magnetization directions that are the same age when compared with the North American APWP, within the limits of the technique. This clearly indicates a common magnetic history for the whole study area. If the remagnetization is a true chemical remanent magnetization (CRM) that was acquired at the stage of folding indicated by the fold tests then the existence of the ESF and VSF groups indicate that two or more closely spaced remagnetization events must have occurred.

At the present a unifying mechanism for all the remagnetization in the Appalachian orogen remains elusive but all proposed remagnetizing mechanisms can be broadly split into two groups: those invoking in situ burial processes and those invoking externally derived fluid migration. In both cases an assumption of simultaneous remagnetization must be accepted for all sites along this fold. For example, the units are stratigraphically adjacent, and all sites are within <300 m stratigraphically but collectively they have been buried by 4.5–5.0 km (Evans & Battles 1999). Thus, it would be difficult to imagine any in situ burial process such as clay alteration or organic matter maturation that was not simultaneous everywhere at our scale of sampling. Also any method for invoking fluid movement would result in simultaneous remagnetization because both units have identical post-depositional histories. Therefore, any mechanism that would induce fluid migration in one unit would have to be active at the same time in the other.

The absence of reversals in the data also raises the question of duration of remagnetization and the possibility that secular variation has not been cancelled out at each site. We note that the timing of remagnetization is within the Kiaman reversed superchron and gives concordant directions to the APWP. Similar magnetizations have been found elsewhere in numerous studies of undeformed rocks and they have all been faithful recorders of the ancient dipole-averaged magnetic field. Secondary magnetizations with very small within-site dispersion, as observed here, have less dispersion than expected from secular variation and therefore they must have been acquired over a long enough time to average out its effects (Van der Voo 1993). Furthermore, Lewchuk & Symons (1995) showed that the bias in their dispersion is similar to the long-term trend of the APWP. This requires that the duration of remagnetization be at least a couple of million years in length, far longer than necessary to account for secular variation.

Regardless of the remagnetizing mechanism, we believe that the acquisition of the magnetization was coincident throughout our study area and thus it cannot be used to explain the ESF and VSF groups.

Effect of post- or syn-magnetization deformation

If we rule out contamination and differences in the timing of acquisition of the remagnetization we must either accept the fold test results as is or question the assumption that the magnetization behaved as a passive marker during post-magnetization deformation. The coincidence of remagnetization with tectonic deformation and the almost ubiquitous occurrence of ‘synfolding magnetizations’ has led many researchers to propose that the origin of this widespread remagnetization is the product of strain or stress related to the deformation (e.g. Hudson et al. 1989), especially in highly deformed rocks.

For example, magnetization with synfolding characteristics can be achieved by mechanically rotating grains during deformation (e.g. Kligfield et al. 1983; Hirt et al. 1986; Cogne & Perroud 1987; Van der Pluijm 1987; Kodama 1988; Stamatakos & Kodama 1991a,b). This process can lead to an ‘apparent synfolding’ result where the new magnetic directions would not relate to the ambient field at the time of deformation, but to the sense of rotation. Kodama (1988) modelled the effects of strain on simple folding and concluded that, shear strain associated with flexural flow folding can cause a pre-folding magnetization to appear to be a synfolding result, but the observed strain in Palaeozoic sedimentary rocks was insufficient to account for the distribution of magnetization directions relative to folding. Stamatakos & Kodama (1991a) argued that bedding parallel shear had reoriented a pre-folding magnetization in the Mississippian Mauch Chunk Formation to produce its synfolding result while Kent & Opdyke (1985) argued that incomplete removal of a secondary magnetization was responsible. Stamatakos & Kodama (1991b) also suggested that grain-scale deformation may have also caused a pre-folding, Silurian magnetization in the Bloomsburg Formation to appear as the synfolding Devonian magnetization as previously reported by Kent (1988). Thus, the role of strain must be considered.

23 sites from this study have both palaeomagnetic and strain partitioning data of which 15 fall into the ESF group (Fig. 7, top) and eight into the VSF group (Fig. 7, bottom). When we compare the results of the palaeomagnetic and strain partitioning data several patterns emerge. Four sites exhibited GBS (18, 36, 44 and 45) and although all four are in the ESF group there is not enough data here to draw conclusions. 12 sites had evidence of TW and they are split with seven in the ESF group and five in the VSF group. However, a relationship does exist between the pressure solution and remagnetization (Fig. 8). With the exception of one anomalous site (68) for which we have no explanation at present, the six sites with the highest combined CPS+TPS (17, 35, 46, 53, 66 and 67) all retain VSF remagnetizations and the average CPS+TPS strain of all eight VSF sites is 21.1 ± 8.2 per cent (1σ). The eight sites with the lowest combined CPS+TPS (18, 36, 37, 39, 44, 45, 69 and 70) retain ESF magnetizations and the average CPS+TPS strain of the 14 ESF sites (excluding 68) is 8.8 ± 6.7 per cent (1σ). Thus, the sites with ESF magnetizations generally have lower CPS and TPS strain.

Figure 8.

Total pressure solution versus optimal differential unfolding level. The solid line is a linear fit to the data (correlation coefficient =−0.66), excluding the anomalous site 68 (open circle).

We can suggest several mechanisms by which the pressure solution strain may affect the unfolding behaviour. First, the pressure solution may destroy remanence-carrying grains by dissolution, thereby reducing the original magnetic intensity. If the pressure solution is greater on the southeast limb, then a greater fraction of the original magnetic grains would be destroyed. Secondly, the pressure solution would create internal grain stresses within the magnetic grains, which in turn may move dislocations and domain boundaries during folding, thereby altering the original remanence directions. Again, there would be more altered magnetic grains on the southeastern limb. Thirdly, a pressure solution of iron oxide phases would provide iron in solution, resulting in the potential precipitation of new iron oxide grains. These new iron oxide grains will contain remanence directions that reflect the time of grain formation, in this case, during folding. Depending on the abundance of new grains, this may either alter the whole rock remanence to a value intermediate between the original and the new directions, or may completely overwhelm the primary remanence direction. Finally, remanence-carrying iron oxide grains may rotate in the pressure solution zones.

Dissolution and reprecipitation of magnetite during folding could produce a ‘true synfolding’ result where the actual time of remagnetization coincides with the optimum unfolding level. Mechanical grain rotation of a pre-existing remagnetization would produce a ‘false synfolding’ result of a magnetization acquired at some earlier time than the fold tests indicate.

Effect of geography, geometry or scale

Several other relationships exist that indirectly may be a function of strain. For example, of the 11 sites with VSF magnetizations, seven (17, 46, 53, 64, 66, 67 and 98) are on the southeast limb, three (35, 71, 72) are on southeast-dipping minor folds within the overall northwest limb and the remaining site (34) is also on a minor fold. Thus, it is possible that the relationship between strain and preservation of the remagnetization may vary with the scale of folding and/or the geometry of the limb.

A relationship also appears to exist between the dip angle and preservation of the magnetization. The ESF group has a lower mean dip angle than the VSF group (32° versus 41°). As the dip angle increases the effects of CPS and TPS will change. As the inclination increases so does the potential for TPS to be accommodated by existing stylolites generated by CPS rather than new stylolites at oblique angles to bedding. Similarly, if compaction proceeds during the folding process, favourably oriented early CPS zones would be used to accommodate syn-folding compaction strain.

While these are clearly possible mechanisms to explain the variation in our data, the relationship between strain level, strain type and reorientation of magnetic grains needs further study.

Implications for regional trends

The fold test results have implications for the hypotheses of regional trends in remagnetization ages. Miller & Kent (1988) showed a variation in the timing of remagnetization across the Appalachians. Stamatakos et al. (1996) proposed that there is a trend from pre-folding magnetizations in the hinterland, to synfolding magnetizations in the central part of the orogen and post-folding magnetizations in the foreland of the Appalachians. Enkin et al. (1997, 2000) have reported a change from normal polarity magnetization in the centre of the western Canadian Rocky Mountains to reversed polarity in the foreland. Both suggested that there was a temporal trend in remagnetization age with progression perpendicular to the trend of the orogen. Our results indicate that fold test results can vary at the site level owing to local effects within a single structure. Thus, inferences drawn on regional geographic distributions of folding behaviour (e.g. Stamatakos et al. 1996; Enkin et al. 2000) may be suspect. While temporal trends may exist, caution should be exercised when they are defined using differing results of fold tests.

Implications for fold tests

Considerable space in the literature has been devoted to the discussion of the appropriate statistic for conducting fold tests. Our research indicates that any of the statistics currently proposed are capable of yielding the correct answer given the appropriate conditions. Both the M&J and W&E tests work in most instances but the M&J test breaks down for low N, while the W&E breaks down when κ is too large. BV statistics while not used traditionally, can yield valuable information concerning the nature of the underlying distribution. Thus, it may be the most useful of all three fold tests and should be used in conjunction with the other tests. Finally, once the data are closely examined to ensure that a single group of data is being analysed with a fold test rather than two or more subgroups, any of the tests available will yield similar results and these results will stand up to statistical rigour. We suggest here that the need to better understand the underlying assumptions concerning the relationship between remagnetization and deformation is more important than developing new fold tests.

Mechanism for remagnetization

Unfortunately, direct evidence of a simple single mechanism for remagnetization remains elusive. Some of the more popularly evoked mechanisms cannot be ruled out but are difficult to apply. For example, exotic fluids are one of the most common mechanisms suggested for remagnetization in sedimentary rocks, yet the Tonoloway Formation and Helderberg Group lack geochemical evidence for exotic fluids (Evans & Battles 1999; Elmore et al. 2001). Clay alteration, another mechanism previously proposed (Katz et al. 2000), would require an explanation for the ubiquitous but variable coincidence of clay alteration with deformation throughout the Appalachians. A similar problem exists with hypotheses involving hydrocarbon maturation or dolomitization. A burial diagenetic process is most likely the explanation for regional remagnetization, however, an explanation for the mechanism responsible for the variation in synfolding results is lacking. One possibility is that multiple mechanisms are involved. For example, the rocks could retain a secondary but pre-folding CRM that was variably modified by strain during deformation.

Piezioremnant magnetization (PRM) owing to deformation should, theoretically, require much more strain than these rocks exhibit, be partially reversible and in any event have the greatest effect on the low coercivity phases. Thus, strain-induced remagnetizations have often been discounted in rocks lacking evidence for significant strain.

However, true PRM occurs at the atomic scale by displacement of walls or rotation of previously pinned domains under stress in an external field (Dunlop & Özdemir 1997). This is a separate mechanical effect from grain rotation under stress. Here we did observe a relationship between CPS+TPS and remagnetization relative to deformation (Fig. 8). Perhaps the strain accommodating mechanism is of greater importance than the overall strain level and that mechanisms other then PRM must be considered.

It is worthy of note that of all the processes that strain accommodation is subdivided into, only CPS and TPS involve dissolution and reprecipitation, mechanisms that one would expect to involve remagnetization. Strain accommodation by pressure solution may be capable of altering the magnetizations in two ways. Dissolution of iron-bearing minerals and precipitation magnetite from the liberated Fe during the deformation process or mechanical reorientation of magnetite grains as other minerals around them are dissolved.

Evans et al. (2002) have observed concentrations of magnetite in stylolite zones but it is unclear whether this represents new mineral growth or merely accumulation of insoluble minerals. Although the stylolite zone would be the likely source of the dissolved iron-bearing minerals, the Fe in solution could migrate into the matrix and precipitate magnetite in a more dispersed manner and thus not be observable.

Chemical reprecipitation would produce a true synfolding result as new magnetic material would be generated during folding. Mechanical reorientation involving the rotation of pre-existing magnetite grains during deformation would produce a false synfolding result. In this scenario, rocks that behaved as a rigid body during deformation would retain their pre-folding orientation, while the magnetic material in other rocks reacted to the deformation in a more fluid-like manner to produce variable apparent synfolding magnetizations. Thus, in some cases the macroscopic rotation of the beds that we observe and measure in the field resulted in less deflection of the magnetization, yielding a false synfolding result (i.e. it indicated a synfolding magnetization for a magnetization that was acquired prior to folding). Whether the pressure solution is directly controlling the magnetization behaviour by altering the magnetization within the stylolite zones or is simply an indicator of the bulk rock deformation behaviour remains untested. The connection between remagnetization and the pressure solution requires further investigation.

Both of the above two processes require that the rocks already be remagnetized shortly before the initiation of deformation because even the sites with pre-folding magnetizations and no evidence of strain retain secondary magnetizations.

Conclusions

We observed a secondary magnetization that had a varying relationship with respect to the timing of deformation. Several sites had pre-folding to early synfolding characteristics, while others had a more variable response to deformation. The sites with early synfolding characteristics probably retain pre-folding CRMs that are related to chemical remagnetization during burial but prior to or at the beginning of deformation. Those sites that appeared to be the closest to post-folding remagnetizations were also those that exhibited the highest-pressure solution strain and thus probably had their secondary CRM altered by the deformation process. Thus, the relationship between deformation, remagnetization and strain mechanisms plays an important role in the outcome of fold tests. While many strain mechanisms may alter the rock, our data show that high levels of CPS and TPS strain are correlated to the presence of synfolding magnetizations. This suggests that some mechanism (mechanical or chemical) related to the pressure solution process is connected to synfolding remagnetizations. Growth of new magnetite or remagnetization of existing magnetites within the stylolite zones may be particularly important. Alternatively, the mechanical rotation of pre-existing magnetite may be responsible for these results.

While fold tests may yield ‘statistically’ valid numbers the results do not guarantee that the data were originally drawn from a Fisherian distribution or that further deformation resulted in simple, passive deflection of the magnetization. This is demonstrated here by the shape differences in the ESF and VSF groups. The former pass the test for circular symmetry at optimum unfolding but the latter fail, indicating that the fold test was inappropriate for the data set.

Several hypotheses for the origin of the remagnetization are possible, but our favoured hypothesis is that the timing of remagnetization is essentially pre-folding, acquired slightly before or during the onset deformation and that the VSF group is the product of a differential response to deformation. The advantage of this hypothesis is that it explains the variation within the whole palaeomagnetic data set, requiring neither temporal differences in the ages of the magnetizations in these two units nor ignoring the shape of the resulting distributions. It also explains why the ESF group is a better fit to the APWP than the VSF group.

One thing that is certain is that there is a variation in the magnetization characteristics across this single fold even though all sites appeared to have identical demagnetization patterns and directions. Thus, generalizations from one fold to another, based on the results of fold tests, must be viewed with caution.

Acknowledgments

This project was funded by NSF grant EAR-9814913 to R. Douglas Elmore and Mark A. Evans. Raleigh Blumstein, Stacia Canaday, Jan Conder and Angela Miller aided in the laboratory measurements. The satellite image in Fig. 1 was kindly supplied by Space Imaging.