Abstract
Postmortem body curvature introduces error in fish morphometric data. Compared to living fish, the causes of such body curvature in fossils may be due to additive taphonomic processes that have been widely studied. However, a protocol that helps to correct its effect upon morphometric data remains unexplored. Here, we test two different mathematical approaches (multivariate regression and the so-called 'unbending functions') available to tackle fish geometric morphometric data in two exceptionally preserved gonorynchiformes fossil fishes, Rubiesichthys gregalis and Gordichthys conquensis, from the Las Hoyas deposits (Early Cretaceous, Spain). Although both methods successfully correct body curvature (i.e., removing misleading geometric variation), our results show that traditional approaches applied in living fishes might not be appropriate to fossil ones, because of the additional anatomical alterations. Namely, the best result for 2D fossil fishes is achieved by correcting the arching of the specimens (mathematically "unbending" them). Ultimately, the effect of body curvature on morphometric data is largely taxon independent and morphological diversity mitigates its effect, but size is an important factor to take into account (because larger individuals tend to be less curved).