We appreciate the opportunity to discuss our paper on the Cenozoic mafic magmatism of eastern Australia and potential relationships to plate motion and are thankful to Musgrave and Schmidt (2019) for bringing the uncertainties and inconsistencies of Jones et al. (2017) to our attention. The comments from Musgrave and Schmidt (2019) focus on the paleomagnetic record of Australian plate motion, so, to begin, we note that Jones et al. (2017) did not include new paleomagnetic data. Instead, the paper presented new 40Ar/39Ar results from east Australian mafic rocks and evaluated them in light of tectonic reconstruction circuits compiled by Seton et al. (2012). To reiterate the original motivation behind our paper, we created animations from previous tectonic reconstructions to examine spatial-temporal relationships between east Australian magmatism and tectonic processes such as the opening of the Tasman and Coral Seas and collision of the Ontong Java Plateau (OJP) with the Solomon Islands. In particular, we examined (1) the age and location of mafic magmatism in eastern Australia with respect to tectonic reshuffling in the southwest pacific; (2) possible formation mechanisms of magmatic provinces in eastern Australia; and (3) the relationship between the latitude of the Australian plate according to a global moving hotspot reference frame (GMHRF; Doubrovine et al., 2012) and different apparent polar wander paths (APWPs) for Australia using previously published paleomagnetic data from Idnurm (1985), Embleton (1981), and Embleton and McElhinny (1982).
Below we clarify our methodology in compiling these previous results, which, as shown by Musgrave and Schmidt (2019), was poorly explained. Some confusion may have arisen from references to “linear and longitudinal apparent polar wander paths (APWPs)” (Jones et al., 2017, p. 474) versus “linear and longitudinal reconstructions” (Jones et al., 2017, p. 466, 477–478). The latter refers only to the animations, which are either based on a modified rotation file of Seton et al. (2012) or on the same reconstruction tree rearranged around paleomagnetic data from Embleton (1981) and Embleton and McElhinny (1982), as described in the methodology section of our paper. Inferences between plate motion and the timing of magmatism or collisions were based solely on the first animation, which used a base reconstruction file, digitized spreading ridges, and plate boundaries compiled by Seton et al. (2012), modified to include an approximate reconstruction of collisional terrains in Papuan New Guinea (PNG), the docking of the OJP, continent polygons, and several digitized features (i.e., east Australian volcanoes, OJP, and the PNG terrains) linked to the base rotation of their respective plate IDs. The Australia-Antarctica plate circuit was selected to represent models that reflect “linear” motion, as it produces motion paths similar to those of Musgrave (1989), Idnurm (1985), and Torsvik et al. (2008) with well-defined longitudinal constraints. In the case of references to the linear and longitudinal APWPs, they are only compared directly to each other, or to the GMHRF. The comparison was made between the paths to determine which data, if any, in either path was more representative of Australian Cenozoic plate motion.
It may certainly be argued that many of the Australian Cenozoic APWPs cited in our paper are out of date and were perhaps not optimal for our analysis, although little new paleomagnetic data from east Australian Cenozoic mafic rocks have been generated since the 1990s. To address this issue, we present here comparisons based on more recent data (Fig. 1A): a reconstruction tree arranged around the GMHRF and the global APWPs of Besse and Courtillot (2002) and Torsvik et al. (2012). All reconstruction trees require reorganization around Australia to prevent the violation of plate boundaries, as no models are entirely consistent with the previous plate circuits. The animations of the APWPs are restricted to the last 60 Ma of motion. Longitudes have not been altered for the APWPs and, as before, they do not represent the true longitude of the plate. The GMHRF produces almost completely northward motion of Australia (similar to the original “linear” reconstruction), although the plate is generally at higher latitudes for any given time in the GMHRF (Animation 1). Neither the global APWP of Besse and Courtillot (2002) nor that of Torsvik et al. (2012) have poles that are the same latitude as the GMHRF (Fig. 1A), possibly due to the inclusion of poles with poor age constraints in the APWPs (e.g., sediments and weathered horizons).
The global APWP of Besse and Courtillot (2002) is, in some ways, similar to the “longitudinal” Australian APWPs discussed in Jones et al. (2017), in that they both show a period of westward divergence of the Australian plate between 25 and 20 Ma (Fig. 1A). However, plate speeds are substantially different in the Besse and Courtillot (2002) global APWP (Fig. 1B; Animation 2) from both the GMHRF (Fig. 1B; Animation 1) and the older APWPs (Fig. 1C). Northward plate velocity between 60 and 50 Ma was relatively slow (∼30 mm/yr), similar to the GMHRF. Between 50 Ma and 40 Ma, plate speed increased (∼50 mm/yr), but the plate was nearly stationary between 40 and 30 Ma (∼8 mm/yr). Plate motion progressively increased between 30 and 20 Ma, from ∼55 mm/yr between 30 and 25 Ma to ∼100 mm/yr between 25 and 20 Ma, before slowing dramatically from 20 to 10 Ma (11 mm/yr). The Besse and Courtillot (2002) global APWP does not reproduce plate speeds from spreading centers for Australia after 50 Ma, largely due to the lack of paleomagnetic data at 40 Ma, 30 Ma, and 10 Ma and an abundance of data from 15 Ma, which skews the poles. The “longitudinal” reconstruction also indicated a period of slow plate velocity, although the timing is restricted (ca. 25–20 Ma) and agrees with geochronology from age-progressive volcanic tracks in eastern Australia (Knesel et al. 2008). The Besse and Courtillot (2002) global APWP, like the APWPs of Embleton (1981) and Embleton and McElhinny (1982), suggests a period of westward plate divergence (Fig. 1A) that correlates, or slightly predates, a bend in east Australian on- and off-shore volcanic tracks that has been correlated to a period of reduced northward migration and decreased half-ridge spreading rates between Australia and Antarctica (Knesel et al., 2008).
Northward velocities using the Torsvik et al. (2012) model are similar to both the GMHRF and the older APWPs from Australia (Fig. 1A–1C; Animation 3). During the period from 60 to 50 Ma, plate speeds are similar between the Torsvik et al. (2012) global APWP (∼30 mm/yr) and the GMHRF (∼30 mm/yr; Fig. 1B). However, the Torsvik et al. (2012) data indicate a period of slow northward velocity (2 mm/yr) between 50 and 40 Ma that is not reflected in either the GMHRF (Doubrovine et al., 2012), the finite rotation parameters compiled by Seton et al. (2012), or any of the older APWPs from Australia. Between 40 and 30 Ma, plate velocity increases to ∼35 mm/yr. Northward velocity is approximately the same as the GMHRF between 30 and 20 Ma (∼55 mm/yr) but reduces again to ∼35 mm/yr between 20 and 10 Ma. Note that the Torsvik et al. (2012) animation does not go to the origin due to the inclusion of the “0 Ma” point of the track. This point was not included in the Besse and Courtillot (2002) animations. The Torsvik et al. (2012) global APWP broadly addresses the issues inherent in Besse and Courtillot (2002) due to greater availability of data from ca. 40 Ma, 30 Ma, 20 Ma, and 10 Ma. However, while more data are available for ca. 40 Ma, these data are restricted to the North American and African continents and are widely dispersed. The dispersion may explain why the averaged windows suggest very slow plate velocity between 50 and 40 Ma. Furthermore, the Torsvik et al. (2012) global APWP uses only two Cenozoic data points from Australia due to the exclusion of data east of the Tasman line. Nevertheless, the Torsvik et al. (2012) global APWP indicates a period of slower northward velocity in the late Oligocene to early Miocene, similar to the longitudinal APWPs, but slightly postdating the bend in the east Australian seamount track.
Musgrave and Schmidt (2019) commented primarily on how our reconstructions were determined, so below we clarify our methodology in compiling and interpreting previous results. When directly comparing APWPs and the GMHRF, reconstructed points were rotated using poles derived from the longitudinal APWP of Embleton and McElhinny (1982) and the “linear” APWP of Idnurm (1985). Because error margins are unavailable for the poles of Embleton and McElhinny (1982), the longitudinal reconstruction also utilized a combination of the smoothed bands of Embleton (1981), as described in the methodology of Jones et al. (2017). To determine poles, we adopted a method similar to that proposed by Embleton (1981), whereby the ages of poles that satisfy the criteria of Besse and Courtillot (2002) were used with a sliding time window to calibrate the approximate age. We did not use the time-averaged poles of Embleton and McElhinny (1982) because they are over-averaged. For example, the 25 Ma pole combines >20 My of geologic time. Poles selected were 4.5 Ma (Rahman, 1971), 14 Ma (Schmidt et al., 1976), 17.5 Ma (Wellman et al., 1969), 22 Ma (Wellman, 1975), 30 Ma (McElhinny et al., 1974), 37 Ma (Idnurm, 1994), 50 Ma (Embleton and McElhinny, 1982), and 60 Ma (Schmidt and Ollier, 1988). The calibration poles used a sliding window calculation in 10 My bands at 5 My intervals, as in Besse and Courtillot (2002), and latitudes were used to determine the bands. Individual poles used for the reconstruction are from the latitude slices of Embleton (1981) and Embleton and McElhinny (1982), and the ages of the digitally filtered latitude-smoothed poles were correlated to the latitude band (∼5°) of the more constrained ages of sliding-window averaged dated profiles (Besse and Courtillot, 2002), such that they would recreate points along the path. The age determination method of Musgrave (1989) was discarded by Idnurm (1990), and, for that reason, we did not use it. Another method would be to directly fit individual poles to the APWP (as is conventional with paleomagnetic “dating”) and select the nearest points as the best estimate for the position at that time.
Musgrave and Schmidt (2019) also pointed out two figures that contain errors, which we here correct. The errors in the figures are primarily due to misaligned overlays that occurred during the transfer from the reconstructions to the figures. A procedure similar to that used in the construction of figures 11 and 13 of Jones et al. (2017) was used to determine the location of the GMHRF reference poles in figure 10 of Jones et al. (2017), but with a reversed angle of rotation. Finite rotation poles were used in GPLATES (www.gplates.org) to rotate the south pole (90°S, 90°E) and determine the location of the point. Points were then individually overlain as projections onto the base figure, after Musgrave (1989). This method introduced a degree of error that may explain some of the variation between our results and those of Musgrave and Schmidt (2019), and in some cases the projections appear to have been misaligned. Figure 11A in Jones et al. (2017) has an error in the location of the longitudinal 50 Ma point, and we thank Musgrave and Schmidt (2019) for drawing this to our attention. Figure 2 shows the corrected position of the 40 and 50 Ma poles. However, as our modeling and calculations are based on the results of the finite rotation, they are not affected by the discrepancy, and the comparison between the poles of the GMHRF still agree with the 50 and 25 Ma zones of the Embleton and McElhinny (1982) APWP.