• Ultraviolet Photography
  •  

Flower Groups from a Chittka Paper


13 replies to this topic

#1 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 9,115 posts
  • Location: UVP Western Division, Santa Fe, New Mexico

Posted 23 June 2015 - 06:36

Ultraviolet As a Component of Flower Reflections and the Colour Perception of Hymenoptera
by L. Chittka, A. Shmida, N. Troje, R Menzel
12 August 1993
Vision Research 34:1489-1508

573 flower species (1063 "petals") were measured and grouped.
Orange, brown or near black flowers were not included.

Notation:
+ = reflecting
- = absorbing
/ = intermediate (i.e., either moderately reflecting or moderately absorbing)

For easier grouping:
Blue = 400-500 nm
Green = 500-600 nm
Red = 600-700 nm

-UV-B+G+R [16.6%]
UV-absorbing
Human: yellow
Bee: green
Absorbent between 300-480 nm, sharp step @520 nm, reflects beyond that.
Example: Lotus corniculatus

+UV-B+G+R [13.1%]
UV-reflecting
Human: yellow
Bee: UV-green
Example: Senecio verralis

-UV+B-G+R & -UV+B /G+R [25.2%]
UV-absorbing
Human: blue with a red component, purple, pink
Bee: blue to cyan
Example: Polemonium caerulum & Knautia arvensis

+UV+B-G+R [10%]
UV-reflecting
Human: violet, purple
Bee: UV-blue
Similar to preceding group but having a conspicuous minimum around 550 nm
and shortwave reflection from 400 down to 370 nm or more.
Example: Pulmonaria obscura

-UV-B-G+R [2.8%]
UV-absorbing
Human: red
Bee: uncoloured, minimal to no excitation of bee visual receptors.
Absorbs between 300-590 nm. Steep slope around 610 - 620 nm.
Usually hummingbird pollinated.
Example: Justicia rizzinii

+UV-B-G+R [1.6%]
UV-reflecting
Human: red
Bee: UV
Example: Papaver rhoeas

-UV+B+G+R [19.7%]
UV-absorbing
Human: white
Bee: cyan
The step between shortwave absorption and longwave reflection occurs around 410 nm.
Example: Berteroa incana

-UV /B+G+R [3.8%]
UV-absorbing
Human: pale yellow, cream
Bee: cyan to green
Shoulder in blue and strong reflection in green & yellow.
Example: Tillandsia incunda

+UV+B+G+R [.9%]
UV-reflecting
Human: white
Bee: uncoloured, almost equal stimulation of all bee receptors, difficult for bees to detect, ["bee-white"?].
UV reflection extends down to 350 or more.
Example: Asphodelus aestivus

/UV /B /G-R [.8%] and 100% of green leaves.
Moderately UV-absorbing/reflecting
Human: green, in humans the R-G opponent channel signals more on the green side.
Bee: uncoloured, no rapid transitions over the bee's visual spectrum, ["bee-grey"?]
Example: Green leaves

[I've forgotten why this doesn't add up to 100%]

**************************

41% of the flowers in the study are
[+UV+B+G+R], [-UV+B+G+R], [-UV-B+G+R], [-UV-B-G+R]
human: white/white/yellow/red
bee: uncoloured/cyan/green/uncoloured

15% of the flowers in the study are
[+UV-B+G+R], [+UV-B-G+R]
human: yellow/red
bee: UV-green/UV

35% of the flowers in the study
human: violet, blue, purple, pink
And have step between 600 - 650 nm and an approximate Gaussian peak between 380 - 450 nm.

There are no pure UV (black) flowers that are UV-reflective [+U-B-G-R].

There are no cyan flowers [+B+G-R], i.e., no flowers absorbing in the red domain.

UV-dark is more common in the flowers than UV-bright.

The UV reflection is always lower than in other parts of the spectrum.

Generally no reflection functions change more than 3 times over the range 300-700 nm.

Sharp changes in spectral reflectances cluster around 400, 500 & 600 nm.

[+UV+B+G+R] white UV-reflective flowers are predominately visited by nocturnal moths
becaiuse these flowers maximize total reflection for those signal receivers
to whom intensity is meaningful.

Bees lack coding channels for intensity. (Back then. Has this changed?)

**************************

UV-Dark
Human <--> Bee
blue, purple, pink <--> blue, cyan
red <--> uncoloured
yellow <--> green
cream <--> cyan
white <--> cyan

UV-Bright
Human <--> Bee
blue, purple, violet <--> UV-blue
red <--> UV
yellow <--> UV-green
white <--> uncoloured
Andrea G. Blum
Often found hanging out with flowers & bees.

#2 msubees

    Zach Huang

  • Members
  • 389 posts
  • Location: Michigan, USA

Posted 23 June 2015 - 16:51

Thank you, Andrea! I need to read this paper. I have "known" Lars (first author) for about a year now on facebook ;)

Here let us see if it makes sense with real photos. I thought all the Potentillas (all assumed to be P. fruticosa) were totally black under UV, but today I increased the exposure quite a bit, and now I saw the white cultivar actually is not as dark as the pink or yellow one.

1. Visual.

Posted Image

2. UV. After adjusting exposure in the old photoshop (I do not see something similar to "level" in Photoninja to adjust the curve), now I see the white is not as dark as the other two.
Posted Image

3. simulated insect vision (S8612 x 2mm + UG5 X 1.5mm). Is the original white now cyan? a hint of cyan, but not very strong. someone like me almost call it white with a tint of blue/green :D
The original pink is more cyan than the white one?
The original yellow is now dark green. so I guess Andrea is always right, and they are also right? :D :D

Posted Image

Edited by msubees, 23 June 2015 - 16:52.


#3 nfoto

    Former Fierce Bear of the North

  • Owner-Administrator
  • 3,256 posts
  • Location: Sørumsand, Norway

Posted 23 June 2015 - 20:48

A general notice is that estimates of proportions within a population demand proper random sampling strategies.

#4 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 9,115 posts
  • Location: UVP Western Division, Santa Fe, New Mexico

Posted 23 June 2015 - 21:20

Zach, these are looking good!

Andrea certainly is not always right!!! I am often wrong. But I try to stay alert and correct the errors. ;)

Let's work it through. But keep in mind that I do not know the actual spectral characteristics of these flowers. I am simply parsing what I see and then referring to those Chittka groups (which very well may have been altered in some ways since the papers after gathering additional data).

yellow Potentilla:
-UV-B+G+R
bee green
The yellow flower foto shows bee green. :D

white Potentilla:
-UV+B+G+R
-UV /B+G+R
bee cyan or bluish-green
The white flower foto shows cyan-ish colour. :D

pink Potentilla:
Pink is a tough one. Is the Pink simply a desaturated magenta? or is it desaturated red?
Here are a couple of samples. Desat magenta on the left & desat red on the right.
I tend to call both these colours pink when seen separately.
Attached Image: Pinks.jpg
So we would have something like this for the pink flower.
-UV+B /G+R or -UV+B+g+R (I like to use small letters to indicate less reflectivity.)
This would be for the bee between bee blue and bee cyan. More blue than green.
or
-UV /B /G +R or -UV+b+g+R
If it is true that bees do not code for intensity, then the blue & green signals must be strong enough to stimulate the bee's blue & green receptors past some threshold (unknown to me) to get cyan.

The pink flower foto shows cyan-ish colour. :D

So I would say that you pretty much nailed it within the usual constraints we must apply:
1. We cannot be completely accurate or precise with photographic equipment being used outside its specification.
2. Our filters/sensors/lenses/etc have not been measured/calibrated.
3. Software rendering is app dependent. No two converting/editing apps produce the same foto.
4. We are only estimating/guessing the spectral reflectivity of the flowers.
5. Other.
Andrea G. Blum
Often found hanging out with flowers & bees.

#5 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 9,115 posts
  • Location: UVP Western Division, Santa Fe, New Mexico

Posted 23 June 2015 - 21:30

A general notice is that estimates of proportions within a population demand proper random sampling strategies.

Covered under "Other". :D :D ;)
I did not make any notes about their sampling, but I have the paper here on my laptop and will try to look it up later.

I sampled (in another sense) Zach's bee vision foto. Obviously a jpeg has wandered away from the original colour space (and so forth and so on). But I was curious anyway as to how those colours were reading. Here is a little sample (in another sense). On the pink Potentilla I probably selected too bright a spot. This (so-called) "sampling" I did in Photoshop to "measure" the colours was simply done to satisfy my curiosity. I doubt there is anything scientific going on in my sampling strategy. :D :P

In the cyan flowers, there is a little more green than blue.
Ratios are G:B of 1.26::1 for the pink and 1.17::1 for the white. Both these are very close to cyan.

Zach, what you are noticing about the white flower in its bee version is that it is brighter than the pink flowers in their bee version.

Attached Image: DSC01455_v1.jpg
Andrea G. Blum
Often found hanging out with flowers & bees.

#6 nfoto

    Former Fierce Bear of the North

  • Owner-Administrator
  • 3,256 posts
  • Location: Sørumsand, Norway

Posted 23 June 2015 - 23:43

You misread my meaning, Andrea. You referred to the flowers themselves, my comment was to the estimated percentages of colour groups.

Spatial sampling of the flower structure must be pretty difficult to carry out in a strict sense, with flower petals/corollas being so massively spatially variable as demonstrated by photography. Thus one actually (unknowingly?) creates a major aliasing of any pattern, unless spectral data are sampled with a high frequency.

#7 msubees

    Zach Huang

  • Members
  • 389 posts
  • Location: Michigan, USA

Posted 24 June 2015 - 20:11

Does this mean, if we map the colors of the three cultivars in our and bee color space, that they will be much closer in bee's system? They definitely looked more different to our eyes.

#8 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 9,115 posts
  • Location: UVP Western Division, Santa Fe, New Mexico

Posted 24 June 2015 - 22:38

Zach, I don't know. There may be some info in the research papers on bee/insect vision. I've read in one paper that bees have no coding for intensity/brightness. But I'm not sure all researchers agree on that. I have not read enough of the research to take a stand on this. (And, of course, this isn't my field anyway. )
Dig into the research papers and see what you find! Let us know.
Andrea G. Blum
Often found hanging out with flowers & bees.

#9 msubees

    Zach Huang

  • Members
  • 389 posts
  • Location: Michigan, USA

Posted 25 June 2015 - 13:59

I got an impression that they must be closer in bee vision. I think Lars might have published something somewhere.

OK, for application, please guess which one of the following is a red, white and pink begonia. this is in UV. I think now my UV blues will be similar to Andreas :(
Posted Image

#10 nfoto

    Former Fierce Bear of the North

  • Owner-Administrator
  • 3,256 posts
  • Location: Sørumsand, Norway

Posted 25 June 2015 - 15:20

White - red - pink ?

#11 msubees

    Zach Huang

  • Members
  • 389 posts
  • Location: Michigan, USA

Posted 25 June 2015 - 15:57

You got one right! If I tell you which one is right, then you will know the answer to all three. Will wait for one day to post the bee vision one (which will gave more answer, perhaps), then finally the visual one.

It seems there are no hard rules here :(

Last time I did one, also of Begonia, but with much larger flowers, which had 4 colors (a yellow). and I got totally different results:
Posted Image

Posted Image

View Postnfoto, on 25 June 2015 - 15:20, said:

White - red - pink ?


#12 Andrea B.

    Desert Dancer

  • Owner-Administrator
  • 9,115 posts
  • Location: UVP Western Division, Santa Fe, New Mexico

Posted 25 June 2015 - 17:51

It seems there are no hard rules here :(

Creating colour profiles for both Visible and UV will help you attain consistent results.
Andrea G. Blum
Often found hanging out with flowers & bees.

#13 msubees

    Zach Huang

  • Members
  • 389 posts
  • Location: Michigan, USA

Posted 26 June 2015 - 01:46

Yes, profiles will make colors more constant. but I do not think the begonias last year (by G5) and this year (Nex7) are due to camera or processing differences. Mainly due to plant differences, I think.

here is the bee vision version. Now perhaps you know which one is red, at least.

Posted Image

View PostAndrea B., on 25 June 2015 - 17:51, said:

It seems there are no hard rules here :(

Creating colour profiles for both Visible and UV will help you attain consistent results.


#14 msubees

    Zach Huang

  • Members
  • 389 posts
  • Location: Michigan, USA

Posted 26 June 2015 - 16:23

the answer!

happy friday...it would be interesting to see if the two species (?) had totally different pigments (e.g. red in the small begonia was uv reflective, and more absorptive for the larger ones taken last year). artificial breeding might be selecting very different pigments in different varieties...

Posted Image