Short answer
Color vision is not based on a calibration to the sky, vegetation and blood. The current leading theory of the development of trichromatic vision in humans is based on the foraging of fruit in our primate ancestors.
Background
The places of red, green and blue wavelengths in the spectrum are physically defined and, therefore, not arbitrary; see this question on Biology SE.
Before continuing, it is good to mention that trichromacy (the presence of red, green and blue cones in the retina) is typical for primates. For example, dichromacy (<500 and >500 nm cones) is the most common configuration in mammals. The split in red and green photopigment in primates developed relatively late in evolution (30-40 mln years ago), inferring trichromacy in primates from a dichromatic ancestor. Note that many avian and aquatic species may be tetrachromats or up (4 or more photopigments) (Nathans et al. 1999), with the mantis shrimp featuring no less than 10(!) different photopigments (Cronin & Marshall, 1989).
It is believed that the red-green-blue system in primates has evolved due to its benefits for identifying fruit on a background of foliage, and to assess the ripeness of fruits. The food of monkeys typically consists of fruit, and young leaves. Young leaves are light-green, while ripe fruits are often yellow (e.g., bananas) and orange (e.g. oranges, mangos) (Osorio & Vorobyev, 1996). Unripe fruits typically contain chlorophyll and resemble the color of foliage (green). Hence, to discern ripe, nutritious fruits (yellow-orange or 570-620 nm) from foliage (green or 495-570 nm - frequencies mentioned were copied from wikipedia), it makes sense to have good resolution in the red-green area. The red opsin has its maximum at 530, and the green at 560 nm. The blue has its peak a lot lower at 425 nm (Nathans et al. 1999). Hence, the red and green opsins, which evolved late in evolution, are both right in the "ripe fruit area" and have absorption spectra very close together. Hence, there is a lot of color sensitivity (high color resolution) in the red-green area, which covers reds, oranges, yellows and greenish colors. This gives primates a very high sensitivity to recognize fruits, and especially ripe fruits in a green background. The resolution in the green-blue range is smaller (blues, purples).
As a background: color vision in trichromats works by color opponency, and specifically a yellow-blue and red-green color opponency. This means there exists no yellowish blue hue, or a reddish green (De Valois & De Valois, 1993). The red-green opponency means that we have exquisite sensitivity in the red-green pathway, as red cones suppress green cone output and vice versa. In other words, there is a high color acuity in the red-green area that helps to recognize fruits.
As a side note: The trichromat color vision scenario in primates supports the notion that cats (for example), being predators with dichromatic vision, do not benefit from being able to identify ripe and unripe fruits, as mentioned by @GoodGravy. In contrast, cats need excellent night vision and motion sensitivity, being nocturnal hunters. Indeed, color vision versus night vision, and color vision versus high motion sensitivity are both trade-offs that do not go hand in hand (Kelber et al, 2002).
References
- Cronin & Marshall, Nature (1989); 137-40
- De Valois & De Valois, Vis Res (1993); 33(8): 1053-65
- Kelber et al., Biol Rev (2003); 78: 81–118
- Nathans et al., Neuron (1999); 24(2): 299–312
