Comment on "Relative motions of the Pacific, Rivera, North American, and Cocos plates since 0.78 Ma" by Charles DeMets and Douglas S. Wilson

Since Atwater (1970) first proposed the existence of the Rivera (RIV) plate, its motion relative to the adjacent Pacific (PAC), Cocos (COC), and North American (NAM) plates has been controversial, and several widely different plate motion models have been proposed. Two factors for this controversy are clearly apparent, namely, (1) significant recent changes in relative plate motions between the RIV and adjacent plates, and (2) close proximity of the RIV-PAC, RIV-COC, and RIV-NAM Euler poles (i.e., the surface locations of Euler vectors) to the RIV plate and its boundaries. The first factor is most problematic when attempting to derive present-day instantaneous relative plate motion models using finite rotation methods or instantaneous rotation methods. The reason for this is that to produce an accurate present-day Euler vector using these methods, plate motions must have remained constant throughout the Brunhes Chron (i.e., since 0.78 Ma). If plate motions have been changing, these changes will introduce systematic errors into the data. It is well established that statistical methods, such as the finite rotation and instantaneous rotation methods, are not generally useful for removing systematic errors. Thus present-day plate motion models derived from these data, employing either of the two methods, may contain systematic biases if the constant pole assumption is violated. Since the accuracy of the plate motion models depends on how well one can remove systematic errors, the presence of systematic errors may result in inaccurate, although precise, models of present-day motions. One could simply assume, as have DeMets and Wilson [1997] (hereinafter referred to as DW97), that the effects of the recent plate motion changes have not seriously degraded the accuracy of the present-day Euler poles. However, due to the close proximity of the RIV-PAC, RIV-COC, and RIV-NAM Euler poles to the RIV plate, even a small error in the Euler poles can result in substantial errors in the predicted presentday plate velocities along the boundaries of the RIV plate. Given this, one must demonstrate that these changes have not seriously degraded the accuracy of the plate motion models. Without such a demonstration, one must be very cautious in drawing conclusions about present-day plate motions based on either the results of the finite or instantaneous rotation methods. We are in disagreement with several of the assumptions, statements, and conclusions presented in DW97. In the