The Chinese Early Cretaceous bird Confuciusornis sanctus is notable because among the many well-preserved specimens, some exhibit a pair of remarkably long, blade-like tail feathers absent in other specimens. The Peters & Peters (2009) study provides confirmation of the main conclusion of Chiappe et al. (2008), namely that there is no statistical correlation between size (limb bone lengths) and presence/absence of the long tail feathers of C. sanctus. Peters & Peters (2009) also concluded that the two distinct size classes identified by Chiappe et al. (2008)—each containing long-tailed specimens—were the expression of sexual size dimorphism, in which the large class was interpreted as females. We argue that such a conclusion is not substantiated by the available data and that interpretations of the size variability of C. sanctus may involve other biological or taphonomic phenomena.
Chiappe et al. (2008) presented four plausible explanations that, individually or together, may account for the observed size classes within the 106 analysed specimens: (i) two species, (ii) sexual dimorphism, (iii) attritional deaths, and (iv) a particular growth pattern. By adding a few new specimens to the original data, Peters & Peters (2009) embraced only one of these explanations: sexual size dimorphism. Peters & Peters supported their deduction by interpreting what is essentially the same size distribution of Chiappe et al. (2008)—a few smaller specimens and two clusters of larger and distinctly sized individuals—through the lens of considering the growth pattern of C. sanctus as comparable to that of modern birds (individuals reaching adult size within months). This assumption was based on De Ricqlès et al. (2003), who, subjectively using bone depositional rates of 10 µm per day (comparable to rates of some modern birds), deduced that C. sanctus developed adult size in 20 weeks. However, these calculations are questionable. Firstly, given that the cortical bone of C. sanctus consists of different types of tissue deposited at different rates, growth through maturity cannot be accurately represented by a single depositional rate. Secondly, De Ricqlès et al.'s estimations cannot accurately account for the bone formed during the early ontogenetic stages of their studied specimen, which was erased by resorption during medullary expansion. Thirdly, De Ricqlès et al. failed to account for the actual duration of pauses in growth implied by the presence of lines of arrested growth that interrupt the deposition of bone tissue in C. sanctus. Thus, the assumption that growth in C. sanctus was comparable to that of its living counterparts is not well supported. In fact, it is more reasonable to assume that depositional rates in C. sanctus were lower than those of modern birds and that skeletal maturity was reached after several years of growth, a condition inferred for other Mesozoic pre-modern birds (Chinsamy-Turan 2005; Erickson et al. 2009) and determined in some basal living birds (Bourdon et al. 2009). The latter assumption is at least consistent with the interpretation of the size distribution of the sample analysed by Chiappe et al. (2008) as a growth series. The Peters & Peters (2009) hypothesis is additionally weakened by not entirely being supported by their own statistics. In contrast to what was proposed by Chiappe et al. (2008), Peters & Peters (2009) argue that the scaling of the long bones of C. sanctus is not isometric and that the smallest specimens constitute a distinct cluster with wing bones that are relatively longer than in larger individuals (i.e. allometric growth). Such an argument is puzzling, however, given that the confidence intervals computed by Peters & Peters (2009) included the smallest individuals on their lower limit, which indicates that all values can be fitted to a single regression line whose slope is virtually 1 (i.e. reduced major axis on the log-transformed variables of the entire sample fits lines with a slope range of 1.03–1.07). Consequently, the diverging slopes of Peters & Peters (2009, figs 1a,b,c,e) for the smallest individuals in the studied sample are statistically unjustifiable. Thus, their argument that the smallest individuals are the only juveniles in the sample (the two larger clusters corresponding to adults of different sexes) cannot be sustained, let alone their claim that the larger cluster is composed of females. It is worth noting that the second dimension of the principal component analysis of Chiappe et al. (2008) indicates that the variability of limb proportions (wings versus legs) is the same throughout the entire sample. Consequently, outliers of small, middle and large size indicate that examples of individuals with relative longer wing bones are not restricted to the smallest sizes. A greater understanding of the morphological diversity of C. sanctus may be reached using shape analysis (geometric morphometrics), but the taphonomic distortion of all known specimens puts a caveat on this approach.
Given the above, it is evident that Peters & Peters (2009) have provided a rather inadequate solution, compounded by questionable ideas about growth patterns and ad hoc statistics, to a far more complex situation. The exceptional preservation and large sample of available specimens of C. sanctus offer an unprecedented opportunity for studying the variation, development and life history of early Mesozoic birds. The elongated rectrices of this bird may have played a reproductive role, but if so, determining this would likely be masked by factors such as molting and differential preservation (Chiappe et al. 1999). Deciphering this 125-Ma-old conundrum may require interpretations that allow the interplay of multiple biological explanations as proposed by Chiappe et al. (2008).