Response to “The importance of controlling genetic variation – remarks on ‘Appropriate rearing density in domesticated zebrafish to avoid masculinization: links with the stress response’”

We appreciate the comment by Delomas and Dabrowski to the Ribas et al. (2017a) paper and want to mention that we were very much aware of the possible effect of genetic differences in the sex ratio response to increasing rearing density from the beginning. This is why, in each one of the four replications of experiment 1, we avoided using pooled eggs and sperm from different progenitors, and instead used eggs and sperm derived from a single dam and sire, respectively. Thus, each time we used a different pair (pairs 1–4), hence creating four different families (biological replicates). Although some sources (e.g. FishBase) claim higher values, the absolute fecundity of zebrafish females is typically around 300 eggs (Ribas and Piferrer, 2014). This creates an experimental problem when eggs from individual dams are used, as in the Ribas et al. (2017a) paper, because with that number of available eggs it is very difficult to test four rearing densities starting with 25, 50, 100 and 200 fish per tank, as even with only two technical replicates per density the amount of eggs needed climbs to 750. Thus, as Delomas and Dabrowski rightly point out, pairs were not tested at all densities, as we explicitly show in table 1 of Ribas et al. (2017a). In that study, pairs 2, 3 and 4 showed an increase in the number of males, while pair 1 showed the opposite effect but, again, the masculinizing effect of elevated density was seen when the sex ratios obtained at each density were compared with the sex ratios of the lowest density, as also stated. Genotype-by-environment (G×E) interactions in the sex ratio response to external factors in zebrafish have been recently described by Ribas et al. (2017b), in accordance with the existence of the sex ratio variation among, but not within, families of domesticated zebrafish, as previously proposed by Liew et al. (2012). This was reflected by different susceptibilities of elevated temperature among different families (see fig. 1C of Ribas et al., 2017b).

Furthermore, in Ribas et al. (2017a) we clearly stated that ongoing experiments in our laboratory confirm the masculinizing effects of elevated densities. We advance the sex ratio results of these experiments below, which are part of a larger study on other aspects of the effects of rearing density. We reared five additional pairs precisely to better determine the effect of genetic variation in the masculinizing response to elevated rearing density. In order to avoid the shortcomings of not having enough eggs, in this case we used only two densities: 11 and 40 fish l⁻¹ in 2.7-liter tanks for the low density and high density groups, respectively, applied during the 18–45 days post-fertilization (dpf) period. Each density treatment was replicated twice for each family pair. This is possible with the typical ∼300 egg batch. As previously observed in the temperature experiments (Ribas et al., 2017b), we found a G×E interaction in the response to elevated densities as evidenced by non-parallel, family-specific reaction norms (Fig. 1). Resulting sex ratios were analyzed at 90 dpf with the chi-squared test. Of the five families tested, family 2 was excluded from the statistical analysis because of insufficient fish numbers at the time of sampling. Of the remaining four families, families 1 and 3 showed statistical differences (P<0.05 and P<0.01, respectively), whereas for families 4 and 5, although there was also an increase in the number of males, differences were not statistically significant (Fig. 1). If the data for all five families are combined, differences are significant (P=0.0014). These data, along with the data presented in Ribas et al. (2017a), clearly illustrate that at a rearing density of 40 fish l⁻¹ or higher, masculinization occurs in zebrafish. Although most families tend to increase the number of males in response to elevated density, some show statistical significance and some do not. Thus, as stated in the concluding remarks of our initial paper, there is an inter-family variation, meaning that there is, as in many other aspects, a genetic component in the sex ratio response to rearing density.