While perusing the questions in the Mammals forum, I visited a link* provided as an answer to one of the older questions: "Colourblind cats?" The link stated that while cats and dogs see the blue and green wavelengths of the color spectrum but not the red, horses see the blue and red wavelengths but not the green. How do you suppose these different vision spectrums evolved? Could the common ancestor of carnivorans and ungulates have been trichromatic, and each set of descendants secondarily lost a cone type? On the other cand, could the descendants of a monochromate ancestor have each gained a different cone type? Or could there have been a dichromatic ancestor whose descendants lost one type of cone an gained another? (Yes, I know. Another long, complicated question whose answer, in all likelihood, will have to be based on speculation. Sorry.)

*www.ccmr.cornell.edu/education/ask/index.html?quid=165

Apologies for the repetition of some information that I posted in an answer about turtle vision, but the same facts are relevant. The first thing to bearing mind is that although we talk about 'red', 'green' and 'blue' light (and red, green and blue sensitive cells in our retinas), this is a bit misleading. The sensation of colour actually arises from relative degree to which each cone cell type is stimulated by incoming light. So, if we see a red object, it is not because the longwave-sensitive ('red') cone sends a message to the brain, it's because a nerve cell that COMPARES the simulation of the red and green cones send a message to the brain saying 'more longwave than mediumwave light'. The reason for this pre-amble is that it's not that cats see blue and green while horses see blue and red, in fact BOTH will see the world through the same colour palette that allows easy discrimination between objects that vary in relative amount of shortwave to medium-to-long-wave light, but a lack of ability to discriminate medium and longwave light (just like a red-green colour-blind human). So, a cat wouldn't be able to tell the difference between a bright reddish object and a dull green or yellow object, because BOTH these objects would stimulate the cat's single 'green' cone to similar degrees. A horse wouldn't be able to tell the difference between a dull reddish or yellow object and a bright green object, because both these objects would stimulate the horse's single 'red' cone to similar degrees. The key point is that all mammals (except most primates) see the world from a a similar colour palette (which one might call blue to yellow, as long as one bears in mind that, to these animals, green, yellow, orange and red are all much of a muchness). Some mammals have their medium-wavelength-sensitive cone tuned to relatively shorter, some to relatively longer, wavelengths but the overall sensitivity of the different types overlaps considerably. Also, some mammals have their shortwave cone tuned to shorter wavelengths than others (e.g. mice have maximum sensitivity in the UV, not blue at all).  As for the evolution of the different types of mammalian colour vision, well it's certain that the ancestor of placental mammals (i.e. most mammals, but not marsupials or monotremes) was a dichromat. We know this from comparing the genes of the different visual pigments across many types of vertebrate. Some marsupials are trichromats, although to my knowledge it's not yet known whether this represents the ancestral condition of mammals and some marsupials (like placentals) have lost a visual pigment, or whether the few trichromatic marsupials have done something similar to primates and evolved a third visual pigment. Outside the mammals, many vertebrates have not two or three, but FOUR types of cone cell: red, green, blue and one type sensitive to the ultraviolet (wavelengths shorter than blue, to which humans are blind).  So, as well as seeing differences between colours that look identical to us (e.g. with or without UV reflection), they will see hues we cannot imagine. For a bird or turtle to explain to us what their colour world was like would be as difficult as a normal-sighted human explaining the differences between red and green to someone with red-green colour blindness.
It's easy to imagine why excellent colour vision might be useful to a bird (just look at their plumage and the brightly coloured insects they eat or avoid eating). But why do turtles, for example, have such excellent colour vision when all they (apparently) do is drift around the ocean eating jellyfish? No-one knows really, but colour vision is always good for telling objects apart when there is a lot of variation in the amount of light (e.g. dappled shadows and lights from the water's surface) -- the relative amount of different wavelengths stays much more constant even when there's big variation in the absolute amount of light. Colour vision only ceases to be useful when there isn't much light at all - the premium being on getting as much light, any light, as you can, rather than comparing the amount of different wavelengths. It's worth bearing in mind that 'four cone types with UV vision' is the ANCESTRAL state for vertebrates (even the humble goldfish has this type of colour vision) and it's mammals, including humans, that have poor colour vision. The best guess is that, because of a long nocturnal period in their evolution (when the dinosaurs were about), colour vision wasn't so useful at night, so mammals lost two of these ancestral vertebrate visual pigments. Most mammals are thus 'dichromats' (two cone types - short and medium- or long-wave). Primates, probably because of the advantages of detecting red/orange fruit (and some reddish young leaves) against a green background, evolved a third visual pigment, this being what we have inherited. So primates have re-acquired some of the colour vision capacity found in many fish, amphibians, reptiles (including turtles) and birds. This has happened more than once in primate evolution: the trichromatic primates (all Old World, some New World and a few lemurs) have evolved their different versions of trichomacy separately. Again, we know this from analysing the visual pigment genes.