The evolution of red blood cell shape in a continental radiation of fishes

The size and shape of Red Blood Cells (RBC) can provide key information on life history strategies in vertebrates. However, little is known about how RBC shape evolved in response to environmental factors and the role of phylogenetic relationship. Here, we analyzed RBC morphometrics in a continental radiation of fishes testing the hypothesis that phylogenetic relationship determines species occupation of morphospace. We collected blood samples of five specimens of 15 freshwater fish species from six orders and used basic stereological methods to measure cell and nucleus area, perimeter, and diameter, cell and nucleus volume, nucleus:cytoplasm ratio, and shape factor of 50 cells per specimen. Then, we conducted a phylogenetic Principal Components Analysis using a dated phylogeny and built a phylomorphospace. To test if the phylogenetic relationship predicted the phenotypic similarity of species, we calculated multivariate phylogenetic signal. We also estimated the evolution rate of RBC shape for each node and tip using ridge regression. Finally, we tested if the position in the water column influenced RBC shape using a phylogenetic GLS. RBC shape seems to have evolved in a non-stationary way because the distribution pattern of species in the phylomorphospace is independent of the phylogeny. Accordingly, the rate of evolution for shape was highly heterogeneous, with an increase in the genus Pygocentrus. Water column position does not influence RBC shape. In conclusion, RBC shape seem to have evolved in response to multiple selective pressures independent of life history characters.

Supporting material), i.e., 250 cells per species (50 cells * 5 animals). From these linear 1 4 2 measurements, we calculated the following second-order stereological measurements: To calculate length, we took the mean of the largest and smallest diameters. Then, to circularity (or shape factor) were estimated by the formula: where A is cell area and P cell perimeter (Russ and Dehoff 2000). Circularity ranges 1 5 1 from 0 (elliptical shape) to 1 (circular shape). All measurements were made in Motic 1 5 2 Images Plus 2.0 in a double-blind fashion (see Fig. S2).

5 3
We calculated the mean of each variable for each species (Table S1), which were using the corrplot package (Wei and Simko, 2017) of R software. To carry out the comparative analyzes, we first obtained a phylogeny with which we have phenotypic data ( Figure 1). This phylogeny was obtained from a fully- Components, which allows us to explore how lineages occupied the morphospace as phylomorphospace function of the phytools package. Monte Carlo randomization procedure that allows testing the significance of the To test the effect of the position in the water column species usually occupy on 1 9 2 RBC shape, we gathered data on depth range from FishBase (Froese and Pauly 2019) and treated it as a categorical variable, henceforth referred to as habitat. Thus, species were classified as demersal (n=4), pelagic (n=2), and benthopelagic (n=9). Afterwards, we compared the fit of evolutionary models to the eight pPCs. Specifically, we fit 1 9 6 Pagel's lambda transformation, Brownian Motion, and Ornstein-Uhlenbeck models to 1 9 7 data using the leave one out cross-validation of the penalized log-likelihood, as  Finally, to test the hypothesis if a change in evolution rate is the mechanism 2 0 5 allowing species to occupy different positions in the morphospace, we estimated tip-2 0 6 level evolution rate of RBC shape using phylogenetic ridge regression, as implemented pPCs. This analysis also automatically finds nodes in which there was a rate shift. The The shape of Red Blood Cells (RBC) of fish species studied varied from oval to perimeter seem to vary less within species (Table S1).

3 1
The distribution of species in the phylomorphospace suggest that closely related 2 3 2 species do not have RBC with similar shape (Fig. 4). Accordingly, RBC shape does not is a high divergence in RBC shape within Characiformes. Therefore, there seems to be a 2 4 5 variation in the pattern of RBC shape evolution at the level of higher order groups 2 4 6 (Superorders), instead of a homogeneous pattern throughout the whole phylogeny. Habitat did not influence RBC shape (Pillai's trace=0.6936; P=0.461). The results did not change if we use a multivariate PGLS (Table S4). We also found that 2 4 9 evolution rate of RBC shape was very heterogeneous at the tip level (Fig. 5). The of evolution. Therefore, our results suggest that an increase in the rate of evolution does 2 5 7 not necessarily produces morphological specialization and rate disparity is not the main 2 5 8 pattern involved in the formation of RBC morphospace. We found that the shape of Red Blood Cells (RBC) varies greatly among and 2 6 3 within freshwater fish species. Contrarily to our initial hypothesis, we did not find a 2 6 4 correlation between water column position (proxy for oxygen availability) and RBC 2 6 5 shape. Accordingly, we did not find phylogenetic signal in RBC shape, suggesting 2 6 6 closely related species do not have similar RBC shape. There does not seem to be a 2 6 7 single mechanism driving the evolution of RBC shape, since the phylomorphospace 2 6 8 shows a non-stationary pattern in the species similarity.

6 9
Small RBC allow species to occupy environments with low dissolved oxygen,  1979, Lay andBaldwin, 1999). Synbranchus had higher perimeter (i.e., higher 2 7 5 surface) and higher area. Thus, our results suggest that this species could have a lower 2 7 6 metabolic rate than Poecilia, because the later has a smaller cell area and perimeter.

7 7
Most species occupied the center of the morphospace, independent of the 2 7 8 phylogenetic relationship. This clustering of distantly related species suggests that 2 7 9 convergence, apparently not caused by water column position, could be a mechanism are essentially from freshwater, while Acanthopterygii has marine origins, with 2 9 3 secondary freshwater introgressions. One interesting pattern that emerges from the 2 9 4 morphospace is that apparently there is no constrains in the morphospace, because Hyphessobrycon, and had high cell area and perimeter, and low nucleus:cytoplasm ratio, with care, since our sample size is not large.

0 3
We found a high heterogeneity in evolution rate of RBC shape. This high rate 3 0 4 heterogeneity helps explain why we did not detect a significant phylogenetic signal, i.e., 3 0 5 the change in shape seems to vary strongly at the tip level, instead of large clades. This analysed (Felsenstein 1988). The large intraspecific variation in most variables ( Table   3 1 0 S1) may suggest that variables describing RBC shape are under weak selection pressure increase in evolution rate, which allowed them to diverge from their sister species, 3 1 4 remaining Acanthopterygii, Hyphessobrycon, and Metynnis respectively. However, 3 1 5 given our sample size we cannot rule out that RBC shape of these species are evolving Water column position did not explain RBC shape in this freshwater fish species radiation. This result suggest that RBC shape may not be entirely determined by partial 3 2 1 pressure of dissolved oxygen in water. It also suggests that water column position does 3 2 2 not represent a main constraint in the evolution of these traits. A recent study (Minias