Add to the previous reference list a new review:
 
Randolf Menzel & Avi Shmida. (1993). The ecology of flower colours and the
natural vision of insect pollinators: the Israeli flora as a case study.
Biological Reviews 68:81-120.
 
The colours seen by bees differ from those seen by humans, and bees do see
very wellin the uv spectrum (to 300 nm wavelength).  The authors divide the
colour spectrum into 6 main colours that insects can sense from mixtures of
uv (U), green (G) and blue (B) based on the colours detected by their 3
types of light sensitive cells [photoreceptors].  They call these U, U-B,
B, B-G, G, and U-G, depending on the degree of colour mixing as colors
reflect a mixture of lights of different wavelengths.
 
B-G includes white and pink+red flowers as humans sense them
U-G includes many yellow flowers
U   many flowers appear pink+red, or green+cream, or yellow.
 
I still don't quite understand how the light mixture is used to quantify
the amount of colours reflected from one flower, but its basically a plot
of each color (U, B, and G) relative to the other to give an overall
mixture.  At a simplistic level a color reflected as 45% B, 37% U, and 18%
G would appear to be defined a U-B flower with a component of G in it.
Maybe someone can correct me if this is wrong.
 
B-G and U-G flowers are not visited by bees as much as by flies and
beetles.  U-B and B flowers are visited by bees in the main. Pure U and
very pure B flowers were not found.
 
However the authors point out, colour sperctrum of the flower alone isn't
everthing.  It is important for flowers to have colours that will stand out
against a fairly uniform landscape, like a green tree or desert, so they
can attract pollinators.
 
Paul Lehmann
Department of Microbiology
Medical College of Ohio.
[log in to unmask] or [log in to unmask]