What processes constrain and redirect the outcome of the adult phenotype? We found that for cardiac output in larval zebrafish there is family specific developmental response to the environment (Table 3). While phenotypic development is clearly redirected by oxygen stress, this redirection is not consistent between families. These results indicate that genetic background is important in determining the cardiovascular system's developmental response to the environment. It is therefore important to investigate developmental response in multiple genetic backgrounds. From an evolutionary perspective, understanding the interplay between sources of variation in cardiac output response to the environment is critical to determining the degree to which that response is adaptive.
Plastic response to environmental variation is a major topic in evolutionary biology (for a discussion see [3, 21, 22]). In this study, plastic response of cardiac output was family specific, indicating that there is likely to be heritable variation for plastic response in cardiac performance. The significant family level variation in response to hypoxia in this study is the sum of parental environmental sources (e.g. "maternal effects") and genetic sources of variation. Variation in parental environment, however, was minimized due to the use of a common and tightly controlled environment (see methods). For this reason we believe that a large portion of the family level response that is observed in this study represents genetic variation.
Evolution of adaptive responses to variable environments requires heritable variation for plastic response. Given the proper selective environment, the variation in plastic response we find in cardiac output could be the raw material of future adaptation. To understand the dynamics of adaptation of a complex trait however, it is important to understand something about the way in which subsidiary traits contribute to that complex trait. If there are interacting traits that can combine in multiple ways to create optimal or near optimal response, then a rugged adaptive landscape will be produced [23, 24] that will maintain variation by providing multiple genotypic targets for selection [25, 26].
Cardiac output is the product of heart rate and stroke volume. Stroke volume is, in turn, the difference between end systolic and end diastolic volume. The family specific response to hypoxia for cardiac output could either be created by complex family specific interactions of these underlying traits, by family specific response in a single trait, or by a unified family specific response seen across all traits. Interestingly, all of the cardiac performance measures (Table 3) showed significant family specific responses to hypoxia. Furthermore, the family specific responses to hypoxia in heart rate (Figure 3A) and stroke volume (Figure 3C) do not correspond in all families (note opposing responses in family K but not other families). Likewise ESV and EDV show no consistent correspondence in their response to hypoxia. These families therefore clearly appear to be responding in a complex way creating a wide variation in integrated cardiac response. While one family responds to hypoxia with a 3 fold increase in cardiac output, another decreases cardiac output by an equivalent amount, and a third family remains constant across environments (Figure 3D). It is likely that there is a great deal of genetic variation in plasticity of cardiac output. In this case, however, selection on that plasticity should be constrained by the complex trait interaction leading to cardiac output. This is because selection will depend not on a single trait but on the variable interactions between traits [23, 25].
While all cardiac performance traits show family specific responses to the environment, only heart rate and ESV showed a significant trend in the response to hypoxia (Table 3). When heart rate responded to hypoxia stress it was consistently in the direction of decreased contraction frequency (Figure 3A). Although tachycardia is the most commonly reported response to hypoxia, bradycardia has been documented in fish under chronic developmental hypoxia. The trend was towards reduced ESV in hypoxia relative to normoxia (Figure 3). This weak trend is overshadowed by the clear lack of a consistent trend in stroke volume (Figure 3) to which it contributes.
Variation between families in cardiac performance in this study may be the product of both genetic and confounding maternal effects. While maternal effects on development can be important [4, 27], there appear to be genetically predetermined physiological trajectories that embryos generally follow [6, 9]. This study was not designed to distinguish between these two factors. Maternal effects themselves are the product of maternal genotype and environment. In this case, tight control of the parental environment makes it likely that maternal genetics account for a large proportion of any maternal effects that do occur in this study. Genetics, either directly or through maternal genetic influence, are therefore likely to play an important role in the family level variation documented in this study.
One curious finding is that, despite having the statistical power to detect small family differences in RBC velocity, oxygen environment did not influence RBC velocity either generally or in a family specific manner (Table 2). Given that considerable family specific environmental response in cardiac output occurs, it is at first surprising that corresponding changes are not found in RBC velocity. This may be due to an increase in arterial diameters (Table 1, Figure 1) that reduce the peripheral resistance to blood flow. In this case, this relatively small anatomical modification may ameliorate the effects of cardiac output changes on RBC flow rates. In addition, small increases in hematocrit, which was not measured, could increase blood viscosity also reducing red cell velocity. Increase in hematocrit in response to development in chronic hypoxia has been shown in zebrafish . Finally, red blood cell velocity is also directly proportional to blood pressure. Without in vivo measurement of blood pressure, it remains unknown how pressure is altered by either environmental or family differences.