2.3. Palaeomagnetic database
2.3.1. Data from Gondwana
To better understand the discrepancies in APW paths, it is thus necessary to refer back to the original palaeomagnetic data. For this purpose, the I.A.G.A palaeomagnetic database (version 4.4 – Pisarevsky, 2003; first version from McElhinny & Lock, 1996) has been used. The aim of this paragraph is to try to establish what data could represent Gondwana and where do the discrepancies in the dataset come from, but it is not to discuss the quality of every single data of this database.
The first step is to select all available Palaeozoic – Early Mesozoic poles (200 – 545 Ma) for Greater Gondwana, as described in §.1.2.1 (see Chapter 1, and figure 1.4). Poles provided by the database and considered to represent primary magnetisations are selected, and their site locations are plotted on a world map (figure 2.2; same map as figure 1.4). The upper age of 200 Ma is chosen since continental drift from 200 Ma to present can be modelled (see §.2.7, below).
The following step of selection consists of discarding poles coming from the disturbed margins of Gondwana (shown in red in figure 2.2). This geographical selection is justified by the fact that possible rotation in orogenic belts could not be recognised as, precisely, “reference” data from the craton are not established.
Palaeopoles calculated from the selected palaeomagnetic database (Annexe I; and figure 2.2) have been rotated into African coordinates (figure 2.3; rotation parameters of Royer et al., 1992). The problem of their polarity is subjective but in most cases, one common polarity can be easily adopted, and is here chosen to match those used in previously published APW paths (see §.2.1 above). Nevertheless, pole positions are still very scattered. In particular, some Cambrian and Silurian data are widely separated from the mean poles calculated for the same period. They are shown with a red border on the map (figure 2.3). For the sake of objectivity, however, these data are not discarded, because they have the same likelihood to represent the “true” palaeopole for Gondwana as the others. Because their polarity to represent the South Pole is a matter of debate, the two possible polarities for these particular data have been tested for four levels of selection described below.
Chapter 2 – APWP for Gondwana
Figure 2.2: World map showing locations of palaeomagnetic studies in greater Gondwana, which have yielded primary magnetisations. In green, craton; red: orogenic zones, and in orange: controversial regions (same map as in figure 1.4). Mercator projection.
2.3.2. Analysis per period
One of the conclusions that can be drawn from the published APW paths (§.2.1) is that none are clearly superior and that the possibility of using data coming from suspect “orange” regions is of crucial importance. Moreover, their relative proportion is important for Mid Palaeozoic times (figure 2.4).
Is it possible at this level already (by looking at the scale of geological periods) to distinguish differences between data coming from the cratonic “green” and the suspect “orange” regions? As a corollary, is it possible to suspect already terranes rotation?
This can be investigated by looking at the mean pole per period for “green” regions only and for “orange” regions only. Nevertheless, as illustrated by simulated data (figures 2.5), a couple of erroneous poles can easily lead to deviate the mean pole and to obtain a very large cone at the 95% confidence level (here larger than 30°). Such results are usually not considered further as it is regarded as not significant. This is not the case when density contours are used (figure 2.5-B).
• Mean poles approach
Distinguishing data coming from the “green” and the “orange” regions can be illustrated with the example of Devonian data.
Chapter 2 – APWP for Gondwana
Figure 2.3: World map showing poles from the palaeomagnetic database (version 4.4; 2003) retained after the base level selection described below in §.2.4.1 (see list of poles in Annexe I). Crosses: Cambrian poles; stars (5 branches): Ordovician poles; triangles: Silurian poles; squares: Devonian poles; circles: Carboniferous poles; diamonds: Permian poles; stars (6 branches): Triassic and Hettangian (200 – 248 Ma) poles. Green symbols represents data coming from cratonic “green” regions; orange symbols, data from suspect “orange” regions. Symbols with a red contour are plotted twice in regard with their two possible polarities.
Data presented on a reconstruction of Gondwana in African coordinates. Mercator projection.
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Triassic Permian Carboniferous Devonian Silurian Ordovician Cambrian "Green" regions "Orange" regions Figure 2.4: Relative proportion of data coming from suspect “orange” regions only and from cratonic “green” regions only in the selection data given
in table I, Annexe I (NB:
“Triassic” includes the Hettangian: 200 – 248 Ma).
Indeed, the distribution of data from “orange” regions looks, simply visually, slightly more grouped to the south-west than data from the “green” regions (looking only at squares in figure 2.3). These data might not be comparable as they could simply reflect a younger and an
Chapter 2 – APWP for Gondwana older part of the Devonian segment of the APW path. However, this is not the case as the age averages per periode (note again that the “Triassic” actually includes the Hettangian: 200 – 248 Ma) are equal within their standard deviation (figure 2.6). When mean poles coming only from suspect “orange” regions or only from cratonic “green” regions are plotted per period on a map (figure 2.7), the “orange” Devonian pole does indeed plot more to the south-west than the “green” one, but the two cones calculated at the 95% confidence level are overlapping. This difference appears thus to be not statistically significant. On the other hand, the two cones have a large radius (>16,0°). This entails that any conclusion would be rather speculative. Similarly, it would be speculative to draw any conclusion for the Silurian, while confidence cones are extremely large (about 50°, and even >180° for the alternate polarity with data from cratonic “green” regions: blue path).
It must be noticed in addition that mean poles from suspect “orange” data only and cratonic “green” data only for the Permian also look distinct. From a statistical point of view, their confidence cones are not separated but tangent. This problem probably stems from the fact that the “orange” mean pole is calculated from too few and scattered data, but may also reflect problems encounter for the latest Paleozoic reconstruction of Pangea.
Figures 2.5: Comparison of results obtained by mean direction and by density contours on a simulated scattered dataset.
A- Simulated data considered; squares are erroneous data. B- the mean direction (green dot with white cross) is deviated from the mean direction obtained if the two erroneous data are discarded (violet dot with white cross), and the corresponding 95% confidence cone (dashed green) is large (>30°). Conversely, the density contours (orange) agrees well with the mean direction without erroneous data (violet).
B A 266,24 380,34 430,67 474,17 517,68 266,31 318,17 383,29 425,13 472,50 504,55 225,58 321,17 232,30 0 50 100 150 200 250 300 350 400 450 500 550 600 Triassic Permian Carboniferous Devonian Silurian Ordovician Cambrian
Age average (Ma)
"Orange" regions "Green" regions
Figure 2.6: Age averages per period and their standard deviations (error bars) for selected data from “orange” and “green” regions only. It shows that data are comparable per period as they represent the same segment of time on the
APW path (NB: “Triassic”
includes the Hettangian: 200 – 248 Ma).
Chapter 2 – APWP for Gondwana
• Density peak approach
Even if mean poles may be deviated due to scattering, calculation of density contours shows a quite identical peak density for the Permian. Similarly, density peak positions from “green” and from “orange” regions are analogous.
However, the position of density peak is also particularly subject to inaccuracy when the number of data is insufficient, because two data simply closer one another by chance may then form the highest density.
Nevertheless, a good agreement between position of peak density and mean pole could be used as a good indication for the accuracy of the APW path. It is for instance very not the case for the Silurian with both polarities (figure 2.7).
• Conclusion
Hence, it seems that there is no significant difference between the two datasets is visible when viewed on a scale of the geological period.
As expected from the presentation of some published APW paths for Gondwana, major discrepancies in palaeopole positions range from Late Ordovician to Early Carboniferous in age. However, the statistical quality of all these data is not equivalent, and can be evaluated from a relative objective point of view.
Figure 2.7: Mean pole positions per period (hexagones) for data coming only from suspect “orange” regions: orange path (red: alternate polarities); or only from cratonic “green” regions: green path (blue: alternate polarities). Thin lines represent paths calculated from density peaks (squares). Numbers are age averages for each period. Cones given at the
95% confidence level are shown with different colour for each period. Pink: Triassic (including Hettangian; i.e. 200 –
248 Ma); brown: Permian; black: Carboniferous; violet: Devonian; yellow: Silurian (confidence cone shown with dashed line for alternate polarities, but not shown for the blue path as it exceeds 180°); green: Ordovician; turquoise: Cambrian (dashed line: alternate polarities). On a reconstruction of Gondwana in African coordinates; Mercator projection.
Chapter 2 – APWP for Gondwana