Recently, we suggested that the size spectrum of Confuciusornis sanctus is best explained by sexual size dimorphism (SSD) and is compatible with a bird-like life history (Peters & Peters 2009). We find the criticism of Chiappe et al. (2010) interesting, but irrelevant to our main argument.
Let us follow this argument from the beginning. Chiappe et al. (2008) reported a ‘bimodal size distribution’ in C. sanctus and hypothesized that it reflected ‘a mid-development phase of exponential growth that separates earlier and later phases of slower growth’ (figure 1a). Avian allometry suggests a body mass of approximately 700 g for the larger animals (size class sIII; Peters & Peters 2009) which, according to the hypothesis, were the most mature ones. Mature female birds of this weight produce hatchlings of approximately 30 g, which is an order of magnitude below the estimated body mass of the smaller cohort (sII) in the ‘bimodal distribution’. This gap between hatchlings and sII (figure 1a) was unaccounted for by Chiappe et al. (2008), who did not consider the egg or neonate masses implied by their hypothesis.
The gap in the hypothesis has no equivalent in the fossil record. Some specimens fall right between the estimated hatchling mass and sII (Peters & Peters 2009), corroborating independently that hatchlings must have been much smaller than sII. To close this gap in the original hypothesis, one may postulate that C. sanctus had a biphasic growth curve (figure 1b). Alternatively, the size spectrum may be interpreted as resulting from two monophasic growth curves with different asymptotes, as in species exhibiting SSD (figure 1c). Both hypotheses explain the size spectrum, but the former (figure 1b) postulates a novel, unique type of growth curve, whereas the latter (figure 1c) works with growth curves commonly observed. Consequently, we accepted the SSD hypothesis (figure 1c), precisely because it does not rely on ‘questionable ideas about growth patterns’ (Chiappe et al. 2010).
The above is the logical backbone of our argumentation (Peters & Peters 2009). Chiappe et al. (2010) do not comment on it. Instead, they state that we interpreted the evidence ‘through the lens of considering the growth pattern of C. sanctus as comparable to that of modern birds'. This is incorrect; our argument (figure 1) requires no assumption of this type. The SSD hypothesis is the most parsimonious explanation for the observed size spectrum; it also happens to be compatible with bird-like growth. Therefore, as explained by Peters & Peters (2009), the SSD hypothesis resolves the apparent conflict between the size spectrum and bird-like growth that Chiappe et al. (2008) had discussed. This resolution, however, is a consequence, not a premise, of the SSD hypothesis.
Chiappe et al. (2010) reject our alleged ‘claim that the larger cluster (i.e. sIII) is composed of females', thus criticizing us for the opposite of what we actually said (‘it remains unclear whether this dimorphism was male or female biased’; Peters & Peters 2009). This misunderstanding is surprising, given that it is the hypothesis of Chiappe et al. (2008), not ours, that puts all mature animals into size class sIII which, consequently, must include reproductive females. Obviously, we could not deviate from Chiappe et al. (2008) in this point when we evaluated the plausibility of their hypothesis (Peters & Peters 2009; compare second paragraph of this text above).
According to the SSD hypothesis (Peters & Peters 2009), only the smallest birds (sI) were juveniles, while the two larger cohorts (sII, sIII) represented sexually dimorphic adults. Chiappe et al. (2010) argue that this conclusion ‘cannot be sustained’ because the deviations of juveniles from allometry (fig. 1 of Peters & Peters 2009) lack statistical significance. This is misleading; the SSD hypothesis does not require juvenile deviation from allometry (fig. 1). Such deviation is an interesting developmental feature, though, and therefore we mentioned it. In the same context, Chiappe et al. (2010) state that our ‘inadequate solution’ is ‘compounded by … ad hoc statistics’ because ‘the confidence intervals computed by Peters & Peters (2009) included the smallest individuals … which indicates that all values can be fitted to a single regression line’. This argument is puzzling since we neither showed confidence intervals in our figures nor mentioned any in our text.
Elongated tail feathers occur in all size classes of C. sanctus, including the smallest specimens (Peters & Ji 1999). It is worthwhile remembering that Chiappe et al. (2008) rejected SSD as an explanation for the observed size spectrum for this one reason alone, namely the lack of correlation between body size and ‘the most apparent sexual trait’, elongated tail feathers. The assumption that these feathers are a sexual trait, however, is arbitrary (Peters & Peters 2009). Likewise, none of the issues raised by Chiappe et al. (2010) affects the argument on which the SSD hypothesis rests (fig. 1). Thus, the SSD hypothesis remains the most consistent and least speculative interpretation of the currently available evidence.