HOTSPOT REFERENCE FRAMES
Tectonic plates can be reconstructed in an absolute sense (e.g., relative to the mantle) using the geometry and ages of hotspot volcanoes, but the oldest hotspot tracks are only Cretaceous in age. Many hotspot frames have been developed since the first one emerged in the early 1970s (Morgan, 1972a,b), but these typically assumed hotspot fixity and were defined separately for the Pacific and Indo-Atlantic domains. The recognition of hotspot mobility led to the emergence of a new generation hotspot frames, in which motions of hotspots were modelled numerically and attempts were made to fit hotspot tracks globally, linking them though circuits of relative plate motion. These models are referred to as the “global moving hotspot reference frame” (GMHRF) and Steinberger et al. (2004) developed the first GMHRF-1 accounting for mantle flow and the resulting distortion of plume conduits and hotspot motion. This model, which extended back to 83.5 Ma, yielded reasonable fits to the geometries and age progressions of four hotspot tracks linked to the Hawaiian and Louisville hotspots (Pacific), and to the Tristan and Reunion (back to 66 Ma) hotspots in the Atlantic and Indian Oceans, respectively. GMHRF-2 (Torsvik et al., 2008) used the same plate motion chain but the relative plate motions were slightly different and smoothed. GMHRF-2 was also extended from 83.5 Ma to 130 Ma (Africa) and 150 Ma (Pacific), using rotation rates relative to fixed hotspots ─ but importantly ─ the extension was undertaken separately for Africa/Indo-Atlantic (Fig. 1b) and the Pacific (Fig. 1c). GMHRF-3 (Doubrovine et al., 2012) also included a fifth hotspot track, the New England Seamount Chain (Central Atlantic), a revision of the Reunion track, and incorporated a separate extension for the African (using rotation rates relative to moving hotspots) and Pacific Plates from 83.5 Ma to 120 Ma which differed from GMHRF-2.
PALEOMAGNETIC REFERENCE FRAMES
Paleomagnetic data can be used to constrain the paleolatitude and azimuthal orientation of lithospheric blocks, and therefore guiding the reconstruction of tectonic plates relative to Earth’s spin axis for most of Earth’s history, but this method cannot constrain past longitudes. Therefore, the absolute longitudinal position of a plate for ages older than the Cretaceous is unknown. For circumventing this short-coming of the paleomagnetic method, one can estimate the paleo-longitude of a plate that moved the least since the Pangea assemblage (~320 Ma) and whose subsequent relative motion can be easily retraced. This plate can then be considered as the root of the global tectonic plate network, and be used to construct global reference frames. The “zero-longitude” motion for Africa was therefore proposed by Burke & Torsvik (2004) and used for developing hybrid mantle frames combining hotspot and paleomagnetic reconstructions (Torsvik et al., 2008, 2010; Doubrovine et al. 2012, 2016). The first use of this method kept Africa strictly fixed in longitude back to 200 Ma but it is known from hotspot reference frames that Africa did move somewhat in longitude (~5-11° according to different models) since the Late Cretaceous. To build a hybrid reference frame we use a hotspot frame to a certain time where hotspot tracks are non-existent, poorly-constrained, or insufficient (e.g., 120 Ma in Doubrovine et al. 2012; Fig. 1a, Fig. 2). For times older than this time, paleomagnetic reconstructions are used, but a correction in longitude is applied according to the longitude difference between the hotspot and the paleomagnetic reference frame at the transition time.
A paleomagnetic reconstruction for Africa at 120 Ma (Torsvik et al., 2012) is remarkably similar to the moving hotspot reference frame (Doubrovine et al., 2012), apart from a 10.6° shift in longitude (Fig. 2); that value is then used to correct all paleomagnetic reconstructions before 120 Ma assuming zero-longitude motion of Africa. But paleomagnetic reconstructions must be corrected for true polar wander (TPW) because they are constructed relative to the Earth’s spin axis and not relative to the mantle, as in the hotspot reference frames. TPW is an intermittent rotation of the solid Earth relative to its spin axis, and is determined by extracting such coherent rotation of all the continents by using global plate tectonic models in paleomagnetic reference frame (Steinberger & Torsvik, 2008; Torsvik et al., 2012).
HYBRID MANTLE REFERENCE FRAMES
Combining hotspot and TPW-corrected paleomagnetic reconstructions is known as a global hybrid mantle reference frame (Torsvik et al., 2008; Steinberger & Torsvik, 2008) and the one we prefer is detailed in Doubrovine et al. (2016) and Torsvik & Cocks (2017). The default rotation file in GPlates versions 2.0/2.1 (www.gplates.org) also uses a global hybrid mantle reference frame but the plate motion model (Matthews et al., 2016) is based on a modified version of GMHRF-2 (Torsvik et al., 2008) for the past 70 Myrs, which is directly connected to an Indo-Atlantic TPW-corrected paleomagnetic frame at 100 Ma (Torsvik et al., 2012) with a longitude correction of -10° (Müller et al., 2016); the 70-100 Ma interval is therefore linearly interpolated and thus smoothed.
The Pacific Plate can be directly linked through plate-circuits to the Indo-Atlantic only for the last 83.5 Ma (Fig. 1a), but controversies exist in choosing the best pathway. For times older than 83.5 Ma, the Pacific Plate is surrounded by subduction zones that break the chain, and must be referenced directly to the mantle using fixed hotspot schemes (Fig. 1c); but this only works back to ~150 Ma and before that time there are no means to reconstruct the Panthalassic plates in an absolute sense relative to the mantle nor the spin-axis.
If you use our GMHRFs but e.g. alter relative the plate circuits significantly then the GMHRF is invalidated -- and should not be used. We also recommend the use our latest models (i.e. GMHRF-3, Doubrovine et al. 2012 and apparent polar wander paths listed in Torsvik et al. 2012) since we consider them superior to the older ones.