On the inference of red blood cell size from fossils

In the last decades, technical advances related to microstructural studies of fossilized skeletal tissues have opened new and promising avenues of investigation in the field of paleobiology (Houssaye et al., 2021). In particular, bone histoanatomical measurements have been used as proxies to infer physiological features of extinct species (Faure-Brac, 2024; Padian & de Ricqlès, 2020). Huttenlocker and Farmer (2017) proposed to use the size of the vascular canals and of osteocyte lacunae to infer the size of red blood cells (RBC), and the latter can be used to infer the presence of endothermy. Indeed, in endothermic species, RBC are smaller than in ectothermic ones (Snyder & Sheafor, 1999), this reduction being the result of enucleation in mammals (Snyder & Sheafor, 1999) and reduction of genome size in birds (Gregory, 2002). Successively, these osteohistological measures have also been used by other authors to infer RBC size (Cubo et al., 2023, 2025; Faure-Brac et al., 2022; Fleischle et al., 2019). These useful studies, however, share one aspect that has always worried me: measuring and inferring RBC size in terms of diameter or area does not take into account the geometrical and physiological differences between nucleated (i.e., non-mammalian) and enucleated (i.e., mammalian) RBC. In their recent study, Cubo et al. (2025) constructed a model including the harmonic mean of bone vascular canal diameter and RBC sizes from a sample of extant amniotes, and inferred RBC size in the pseudosuchian Pelagosaurus typus. They inferred an RBC width of 8.76 (8.08–9.49) μm and concluded that “this confidence interval excludes the value separating endotherms with small red blood cells from ectotherms with large ones”. Thus, they put a RBC width of 8 μm as a threshold between endothermic and ectothermic species. In reality, there are many endothermic species (both birds and mammals) with RBC width larger than 8 μm (Figure 1 and Table S1). Moreover, some squamates (thus ectothermic) have RBC width below 8 μm (Arikan & Cicek, 2014). Therefore, using a threshold of RBC width in these studies seems unfeasible. Moreover, using RBC width ignores the great differences between volumes (known in hematology as mean corpuscular volume [MCV]): for example, RBC from echidna (Tachyglossus aculeatus) and mallard (Anas platyrhynchos) are both 6.6 μm wide, but their volumes are 67.5 versus 140 fL, that is, μm3 (Table S1). While bone vascular canal diameter can be a physical constraint to the dimensions of nucleated RBC, this seems not to be the case for mammalian RBC. In fact, these are highly deformable, and the larger they are, the more they can deform, so that they all can pass through 3-μm microtubes (Smith et al., 1979), that is, through vessels smaller than the RBC diameter of any mammalian species. In agreement with this, capillaries have equal diameters in all studied mammals, both in cardiac (Sobin & Tremer, 1972) and skeletal muscle (Potter & Groom, 1983). Therefore, bone vascular canal diameter should not be correlated with RBC size, and indeed, data for mammalian species in the study of Huttenlocker and Farmer (2017) seem to confirm this absence of correlation: neither minimum diameter nor harmonic mean of diameters are correlated with RBC width (p = 0.38 and p = 0.36, respectively) or MCV (p = 0.32 and p = 0.097, respectively; Table S2). However, this does not exclude the fact that in the future (with a larger number of species) a correlation can be found, but this would probably be indirect, that is, both variables are linked to a third one, for example, metabolic rates. For nucleated RBC, instead, canal diameter is clearly a physical constraint, as they are less deformable (Gaehtgens et al., 1981) and the nucleus is quite a rigid inclusion. Indeed, beside the study of Huttenlocker and Farmer (2017), Snyder and Sheafor (1999), studying several amphibian and squamate species, demonstrated a correlation between RBC size and muscle capillary diameter. This correlation could still be indirect, that is, canal diameter and RBC size are correlated because both are correlated with metabolic rates, but this is not a problem in the field of paleophysiology. On the contrary, it could be useful if the aim is to infer metabolic rates and/or endothermy. Even if one follows my suggestion to use MCV instead of RBC diameter, still no threshold can be used to discriminate between endothermic and ectothermic species. For example, the lizard Psammodromus algirus has an MCV smaller than the ostrich Struthio camelus, that is, 188.5 versus 259.5 fL (Table S1). However, RBC size becomes much more informative if it is related to the body mass of the species. Beside differences due to endothermy or ectothermy, metabolic rates inside vertebrate classes scale allometrically. In particular, mass-specific basal (or standard) metabolic rate is inversely correlated with body mass. Similarly, as early as the 19th century, Gulliver (1875), measuring RBC diameter, noted that this scales allometrically and this possible correlation has been studied above all in mammals (Udroiu, 2023). Interestingly, Gulliver (1875) noted that large RBCs are characteristic not only of large mammals but also of what he called “primitive” species (xenarthrans, aardvark, marsupials) and which now we know have low metabolic rates. Indeed, I recently demonstrated that, in mammals, the correlation between MCV and body mass is indirect, and is in reality due to the correlation between MCV and metabolic rate (Udroiu, 2023). Actually, this correlation is also indirect and is due to the correlation between MCV and hemoglobin-oxygen affinity (which is usually, but not always, lower in species with high metabolic rates). In species with nucleated RBCs, a certain degree of allometry for MCV seems to exist, but another important variable influencing RBC size is the dimension of the nucleus, linked to genome size (Gregory, 2002). However, genome size is also correlated with metabolic rate (Gregory, 2002). Therefore, in non-mammals as well, RBC size can be informative of metabolic rate if it is related to body mass. For example, if two species have the same body mass, the one with larger RBCs probably has a smaller mass-specific standard metabolic rate. We could use body mass against RBC width (Figure 2a), but using MCV seems decisively better (Figure 2b and Table S3), for the same reason explained before. In this way, it can be clearly viewed that the regression line for birds (Table S3) lies well below other reptiles, an indication of their higher metabolic rates. Mammals show even lower values, obviously because of their enucleated RBCs. Therefore, I suggest for future studies to use osteohistological measures to infer MCV and to use this together with estimated body mass of the fossil in a scatter plot of body mass against MCV. As an example, I used RBC width and MCV from the species of Figure 1 to infer the MCV of Pelagosaurus typus from its RBC width (Figure S1) inferred by Cubo et al. (2025). When plotted against body mass, it clearly lies well above the regression lines of birds (Figure 2b) and outside the confidence interval of the latter (Table S3), suggesting a mass-specific standard metabolic rate more similar to Lepidosauria. This seems to confirm the conclusions of Cubo et al. (2025), without, however, using a threshold for RBC size. My calculations are just an example and have several limitations (small number of species, biased selection of species, etc.) and my MCV value for Pelagosaurus typus is just a rough estimate. My aim is just to draw attention to the issue of how to consider RBC “size.” My critique on the use of a threshold for RBC size does not mean that no threshold can be used at all. The problem with RBC size is that it scales allometrically with body mass. An interesting issue to explore would be the existence of a threshold for “mass-independent” MCV that can discriminate between endothermy and ectothermy. When using metabolic rates to infer endothermy, Cubo et al. (2025) correctly used mass-independent rest metabolic rates (RMR), that is, RMR divided by body mass to the 0.67 power, as this removes the effect of body mass (because RMR scales with body mass with a 0.67 exponent). In order to use a mass-independent MCV (or even mass-independent RBC width), a wide sample of tetrapod species is required, allowing one to find the exponent with which MCV scales with body mass. I am conducting a study on RBC parameters in vertebrates, similar to my previous one in mammals (Udroiu, 2023), which could give such information useful for paleophysiologists. In conclusion, I make three suggestions for studies inferring RBC size from osteohistological measures of fossils: (a) exclude mammals from nonmammalian models; (b) infer MCV instead of RBC width; (c) plot inferred MCV against body mass. In this way, I think the exciting discoveries made by studies like the one by Cubo et al. (2025) could become even more solid. Ion Udroiu: Writing – review and editing; writing – original draft; formal analysis. Data S1. Supporting Information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.