How many among us becomes a guppy addict after being caught by the colors of these swimming gems in any of the fish shops spread all around the world? Possibly you were a young boy, may be a young girl,  and you no longer remember their shape or size, but surely you yet remember their color. 

To finish my participation in this GuppyLabs e-Bulletin project, I will share with you the references I gathered and a few of my own personal thoughts about how guppies produce the colors that caught my yes at first glance almost three decades ago.

The basic mechanism of color production in guppies revolve around the interactions between five chromatophores (Ueshima et al 1998, Ben et al. 2003), out of the six  kind of chromatophores already reported in lower vertebrates. As far as I know there are no scientific reports of the occurrence of  cyanophores (blue chromatophores) in guppies although an unknown kind of  reflective chromatophore was reported in guppies (Nakajima et al., 1999).  

Chromatophores are cells located in the skin of  some vertebrates that enclose cellular organelles (chromatossomes) which are filled with coloring materials. These coloring materials can be classified as: (1) pigments, or biochromes, when they produce extracts which resemble the color they show while in chromatossomes. In such case the color is produced by the physical phenomena of absorption and reflection of the light. (2) Schemochromes, on the other hand, produce extracts which do not resemble the color they produce arising the so called structural colors. Structural colors are produced by interference, scattering and other physical phenomena.  

MELANOPHORES.

Melanophores are cells which enclose organelles called melanossomes which are filled with melanin. Melanin is synthesized from the amino acid tyrosine present in the food. Actually there are at least three form of melanins: pheomelanin, eumelanin and neuromelanin. But only eumelanin occur in fishe's chromatophores (Ito and Wakamatsu, 2003). Some times people make reference to them simply as melanin.  Accordingly to Fox (1976 - p222) melanins can be black, brown, ruddy, tawny or buff colored.

Goodrich et all (1944), were the first to study the melanophores of guppies with different body base colors... "Our chief interest in this study lies in the comparison of the effect produced by the various gene combinations on cells and cell arrangements. The cells studied have been chromatophores - chiefly melanophores. Of these we recognize the following types or categories: (1) The dendritic melanophores. These are exceedingly irregular in form, and much variation is found. Most frequently there are two or three irregular processes proceeding from a relatively small cell body. They are practically always located in the dermis superficial to the scale and frequently may be lined up at the edge of a scale. (2) The corolla melanophores. These are so named because, in their configuration, they often resemble a composite flower with petal-like processes proceeding from a disc-like center  (Sumner and Wells, 1933). These are located in the dermis, beneath the scales. (3) The punctate melanophores. We believe it to be homologous with the corolla type for reasons mentioned later. They are small rounded cells, occasionally extending short processes..."

... "Our description is of the female pattern, or ground pattern, and is based chiefly on the dorsal aspect shortly anterior and posterior to the dorsal fin. There seems to be complete dominance. Variation in the heterozygote has not been observed. The Wild type (BG) female macroscopically is of a gray-brown color and is characterized by a diamond-shaped pattern within the meshes of which may be seen other somewhat more sparsely and irregularly distributed melanophores. On the dorsal aspect the coverage by melanophores is more uniform and the diamond pattern only dimly visible. The diamond pattern is formed of the corolla melanophores, varying from 42 to 119 in diameter when expanded, and are located at the edges of dermal pockets in which the scales are inserted. From this it follows that they are normally observed on the through the next anterior overlaying scale. The irregularly distributed melanophores are of two types - the corolla types, deeply located beneath the scales, and the dendritic form in the superficial dermis. Both types of cell are present at birth. Counts were made of numbers of melanophores visible in a measured area (0.9 sq mm) on various regions of the body, which gives an average of 120.1 per 0.9 sq mm..."

The Golden (Bg) female exhibits a more clear-cut but also more variable patterns than any of the other types. The difference is due the combination of two factors: first, the larger size of both dendritic and corolla type melanophores, the latter varying from 70m  to 210m  in diameter; secondly, the melanophores are more definitely concentrated in the underlying diamond pattern and also above this at the edges of the scales, producing a superimposed scalloped design which tends to accentuate the underlying diamond network. The interstices of the mesh are often clear of melanophores. The pattern may vary considerably in different parts of the body and in different individuals. This is especially noticeable in half-grown fish and gives a coarsely mottled appearance. ... Neither type of melanophore is present at birth but usually appears during the first week thereafter. This is subject to considerable variation. Counts shown only about 50 percent as many melanophores per unit area as in the Wild and Blond.

The Cream female (bg) is a uniform warm yellow color and no chromatophores are macroscopically visible. The melanophores are most frequently few in number and sometimes are almost completely absent in a given area. They consist of small punctate cells  tending to follow the diamond pattern beneath the scales, while a few dendritic superficial cells  are located near the edges of the scales. Both types of cells may be present at birth. Recently we have found in our cultures some individuals having a larger number of cells, which, nevertheless, do not resemble Blond, since the cells are restricted to the edges of scales on the dorsal aspect of the body. Counts are given in table 4, and here the second and third listed cases are this later type. There is also much variability in the different areas of a single fish. The Cream gives a positive reaction to the Dopa test, discussed below, indicating the presence of  "colorless" melanophores and that therefore the fish may be light colored because some disarrangement of normal metabolic processes. The Cream is not a hardy stock and is difficult to maintain in homozygous condition."

Plate 1 in Goodrich et all (1944).

Please, boss, insert here a figure of Plate 1 in Goodrich et all (1944) including foot notes and title.

PLATE 1

EXPLANATION OF PLATE 1
A—D Photomicrographs of cell pattern of the four phenotypes taken on dorsal midline slightly anterior to dorsal fin. X43.
A, Wild; B, Golden; C, Blond; D Cream.
(C and D show clearly both dendritic and punctate melanophores).
E From side of body of Wild to show diamond pattern. X39.
F From dorsal aspect of punctate type to show punctate and dendritic melanophores. X8o.

Table 5 in Goodrich et all (1944).

Nayudu and Hunter (1978) compared the melanophores of wild type to those of guppies carrying Nigrocaudatus II (NiII) gene: "In the study reported here, a number of differences have been found between NiII and wild-type melanophores in various respects. The NiII melanophores are larger, much more numerous, positioned at multiple level in the dermis and between muscle bundles. This last finding explains the macroscopically greyer appearance of fresh muscle of NiII in comparison with that of the wild-type. The wild-type melanophores are smaller and less numerous, and positioned only in the lower level of the dermis in a single layer.

In another paper published a bit later (Nayudu and Hunter, 1979) they went deeper: "Wild-type melanophores are those present universally in guppies. Others forming variable patterns may be superimposed and are called mutant melanophores. 

Figure 1 in Nayudu and Hunter (1979). .Expression of the Melanie Patterns: a) Male Flavus (Fla); b) Female Fla. Expression similar to male; c) Male Pigmentierte Caudalis (Cp); d) Fe male (:p. Expression less elaborate than male; e) Male Nigrocaudatus II (NiII). blackening of entire caudal peduncle, a portion of the tail fin, and the posterior 1/3 of the body, the tail fin in this example also carries Fla 1) Female NiII, not as intense as in the male; no tail pattern.

Table 1 in Nayudu and Hunter ( 1979). .

Cytology and pattern development -

Wild-type tail fin melanophores visible at one week before birth are always associated with the fin rays and are often situated partially between the two halves of a ray. Although it is difficult to measure their size precisely since their largest surface cannot be observed externally, the diameter is approximately 10 to 20 mm in fish at one week before birth, growing to 50 to 60 mm in 2-month-old fish.

The caudal peduncle carries wild-type melanophores of three different types: Large melanophores dorsal ventral and lateral to the spine; corollar melanophores in the dermis forming the characteristic reticulated pattern for whish Poecilia reticulata is named; and dendritic  melanophores in the dermis covering the scales. In fish one week before birth, only the large dorsal, ventral and lateral melanophores are pigmented. These are also show at post birth stages. From 5 weeks onward the dentrites are so thickly spaced that they appear to have fused together to form a thick branches off which smaller dendrites spread. In mature fishes the melanophores are approximately 200 < mm in diameter, about twice the size of corollar melanophores, and are also more densely pigmented.<

Two to 3 weeks after birth a few precorollar melanophores (approximately 50 mm in diameter) appear and are positioned where the diamond pattern will develop. Gradually over the next 4 weeks these cells take on their corollar shape. The change in shape from dendrite to corollar is the result of proliferation of the cellular projections. The cell body of corollar melanophores is about 9 mm across and the total diameter is 50 to 100 mm.

At the age of 3 weeks to one month, the dendritic melanophores in the scales begin to appear. Fig. 3d shows two of these which have established a bridge between their projections. These melanophores are about one half the overall size of the corollar cells, although the cell body is approximately the same in both types.<

The mutant melanophores in the present study are those which appear under the control of one of the supergene complexes present in domestic populations (Yamamoto, 1975). They have easily distinguishable developmental and morphological characteristics. Flavus tail pattern (Fla) melanophores are intermediate in shape between the dendritic and corollar cells of the wild-type and always occurs in characteristic positions on the tail fin. their overall size varies between 30 X 30 mm to 50 X 70 mm. The first pigmentation appears at 2 weeks after birth and consists characteristically of a triangular patch of melanophores in the anterior dorsal corner of the tail fin where it joins the caudal peduncle, and a streak of melanophores along the ventral edge of the fin. The melanophores spread as they multiply and as result the adult pattern (more elaborate in males than in females) encompasses considerable individual variability in expression while still being recognizable. 

Pigmentierte caudalis tail pattern (Cp) is composed of two morphologically distinguishable populations of melanophores. One of corollar type 90 X 90 mm is positioned along the caudal peduncle-tail fin junction and forms the central streak of the adult pattern. The other, bipolar 80 X 20 mm, occupies the dorsal and ventral edges of the fin. The melanophores are first visible at 3 weeks of age in a characteristic pattern, reminiscent of accentuated wild-type melanophores. The pattern develops to the stage shown in Fig. 5d in females. In adult males it elaborates further. Two extreme examples of a phenotypic continuun are shown in which the critical difference is whether the tail fin growth started before or after the Cp edge melanophores had spread around the posterior of the tail fin. The latter alternative effectively close the pattern and inhibits the spread of the central streak melanophores. 

Nigrocaudatus II (NiII), the only mutant body pattern described in this study, is unique in that it is expressed at birth. The melanophores of which it is composed are relatively large, 90 to 100  mm at birth growing to 200 mm diameter in adults, and are similar in appearance to those which occur in the caudal peduncle, dorsal and ventral apices in the wild-type fish.

In the early development of the NiII pattern the dorsal, ventral and lateral spinal melanophores spread from their original positions and cover the diamond pattern melanophores. The large melanophores then spread to generally occupy the caudal peduncle dermis at all depths (Nayudu and Hunter, 1978). Individual melanophores cannot be distinguished at this stage. In addition, further darkening takes place in the male at the initiation of sexual maturity.

Concurrently with the development of NiII on the caudal peduncle, the melanophores spread onto the proximal portion of the tail fin and produce a general darkening of characteristic appearance. In maturing males the black area becomes sharply delineated, but remains diffuse in females until the age of one year or more."

Besides the above cited anatomic mutations some physiological mutations also occur in melanogenesis pathway leading to defective deposition of melanin on the melanophores, e. g.: real red eye albinos, simply called albinos, and  grape/vine red eye albinos,  called lutinos. 

Once or twice I saw postings in guppy forums relating the occurrence of all grey offsprings in F1 of out-crossings between  two different strains of real read eye albino guppies, that made me think that we have at least two different and complementary kinds of melanin defective guppies. 

Another mutation which effects melanophores is the one called pink were melanophores in the peduncle looks defective.

One last point, possibly about melanophores: Did you ever got intrigued by the different shades of grey found among the so called grey/wild strains???   

XANTHOPHORES AND ERYTHROPHORES

These kind of cells impart yellow, orange to red colors to guppies. Like they enclose pigments of the same groups and like they share the same morphology and physiology I prefer simply call all together as xanto-erythrophores.  Xantho-erythrophores enclose 2 types of coloring organelles; (1) Pteridinossones which contain pteridines, sometimes called pterins too; and (2) carotenoid droplets containing carotenoids. Accordingly Fox (1976), p 293, pterins can confer white, yellow, orange or red colors, while carotenoids, p 67, can be "various light, medium or deep shades of yellow, orange, red or violet in color (may be purple, blue, green, grey or brownish if conjugated with proteins."    

"Pteridines are derived from purines, which are synthesized de novo by animals from carbohydrates and proteins (Hurst, 1980)." accordingly  Grether et al. (2001). As far as I know Takeuchi (1975) was the first one to study the pteridinossomes of guppies but I had no access to his article. Again from Grether et al. (2001): "Two types of pigments were present in the integument of the guppies examined. Yellow hydrophobic compounds (carotenoids) were present in both the orange-spot and the non-orange-spot fractions, while red hydrophilic compounds (drosopterins) were present solely in the orange-spot fraction of males.  Coloured pteridines were completely absent from the integument of females. 

...

 The pattern that was produced precisely matched the fingerprint pattern produced by the three most abundant drosopterins in Drosophila eye extract, specifically drosopterin, isodrosopterin and neodrosopterin (Schwinck, 1975). Colourless pteridines were present as well, notably isoxanthopterin and biopterin (and an unidentified pigment, probably ranachrome-3, according Henz et al. (1977)." 

Before you begin think that pteridines do not impart colors to guppies I must say that  Grether et al. (2001) studied wild guppies populations from Trinidad & Tobago and that they did not included skin pieces taken from the head and from the fins in their analysis. 

On the other hand  Ben et al. (2003) reported the presence of two cDNA isoforms for 6-pyruvoyl tetrahydropterin synthase (PTPS) and of cDNA for xanthine dehydrogenase (XDH) in the caudal fins of guppies cultivated in Singapore. PTPS is an enzyme involved  in the pathway of dihydroneopterin triphosphate to 6-pyruvoyl tetrahydropterin. 6-pyruvoyl tetrahydropterin is later converted to 5,6,7,8-Tetrahydrobiopterin (BH4). BH4 has been associated with the  synthesis of pteridine pigments in the goldfish and zebrafish and BH4 enzymatic activity were found to be significantly high in the erythrophores of goldfish. 

"Among the 12 different guppy varieties that were included in the Northern blot analysis, Yellow Tail, Leopard, and Yellow Snakeskin contain xanthophores in decreasing density in the caudal fin, and the PTPS mRNA levels were found to decrease in the same order. Red Tail, Red Snakeskin, and Red Metallic represent guppy varieties with predominantly erythrophores, and PTPS mRNA levels were highly expressed among them. Purple Tail, Diamond, and Blue Metallic have silvery iridescent caudal fins resulting mainly from the presence of iridophores. Purple Tail looks more yellowish than Diamond and Blue Metallic, while Blue Metallic has deeper blue iridescence than Diamond. The PTPS mRNA levels were moderately expressed in these guppy varieties and appeared to be related to the density of the xanthophores. The lowest levels of PTPS mRNA were found in the Blue Tail and 3/4 Black Tail varieties, in which melanophores are abundant. 

Adult females and juveniles of the Doublesword variety, which lack pigment cells on their caudal fins, were used as negative controls. PTPS mRNAs were expressed at very low levels. In contrast, the male Doublesword contains predominantly xanthophores and iridophores in the caudal  fins. We were able to demonstrate that the PTPS mRNAs were very weakly expressed in the female and juvenile Doublesword but highly expressed in the male. Thus, the PTPS mRNA level appeared to be proportional to the density of xanthophores and erythrophores in the caudal fins of the various guppy varieties investigated. 

... 

The finding that PTPS mRNA expression is proportional to the density of xanthophores and erythrophores on the guppy is in line with a study on goldfish (Masada et al., 1990). In this particular study, the activities of GCH and PTPS were found to be significantly higher in erythrophores than in melanophores, and they were barely detectable in nonpigment cells such as fibroblast-like cells. These results implied that there might be significant production of sepiapterin and other pteridines in yellow-red chromatophores of fish relative to other pigment and nonpigment cells. Our results provide evidence to support the hypothesis that a de novo BH4 biosynthetic pathway is involved in pteridine biosynthesis in xanthophores and erythrophores. Since the PTPS expression level in Blue Metallic is similar to that in the colorless female Doublesword, the association of PTPS mRNA levels with iridophores is not evident. The PTPS expression in 3/4 Black and Blue Tail is, however, significantly higher than in the colorless female Doublesword. 

Three possibilities could explain this result. First, 3/4 Black Tail and Blue Tail have xanthophores and erythrophores in the caudal fins but are masked by a thick layer of melanophores. Second, de novo synthesized pteridines could accumulate in the melanophores as a less abundant pigment. This is consistent with the study by Masada et al. (1990), which showed that melanogenesis in goldfish was accompanied by pterinogenesis. Third, BH4 functions as a critical cofactor of phenylalanine hydroxylase, which converts L-phenylalanine to L-tyrosine (Schallreuter et al., 1994a, 1994b), a precursor of melanin synthesis. Thus, PTPS expression in melanophores is necessary to supply a regulatory cofactor for melanogenesis. The homogeneous expression of PTPS in livers, in contrast to its differential expression in caudal fins of guppy varieties, implies that the high expression of PTPS in liver may not be associated with the pigmentation patterns of the caudal fins. It also further suggests that pteridine biosynthesis in xanthophores and erythrophores is not related to that in the liver." (Ben et al. 2003). 

"Xanthine dehydrogenase catalyzes the conversion of hypoxanthine and xanthine to xanthine and uric acid, respectively (Parks and Granger, 1986). It also participates in the catabolic pathway of BH4, including the conversion of 7,8-dihydropterin to 7,8-dihydroxanthopterin, xanthopterin to leucopterin, and pterin to isoxanthopterin and drosopterin (Hille and Nishino, 1995; Blau et al., 1996; Ziegler et al., 2000). In poikilothermal vertebrates and insects, XDH also controls the synthesis of pigments for metallic and yellow or orange colorations (Oliphant and Hudon, 1993).

 ... 

Northern blot analysis indicated that XDH expression was considerably lower than PTPS expression in the guppy caudal fins (Figure 4, B). Like PTPS, the XDH mRNAs appeared to be differentially expressed among the 12 varieties. The caudal fins of Purple Tail and Diamond had higher XDH expression than did Blue Metallic, Blue Tail, and 3/4 Black Tail. Among these varieties, Purple Tail and Diamond have a higher density of iridophores. Among the predominantly red guppy varieties, Red Metallic had the highest level of XDH expression, and again it has the highest density of iridophores. The XDH level in Yellow Tail was higher than that in Leopard and Yellow Snakeskin, although Yellow Snakeskin is the only yellow variety that has a reasonably high density of iridophores. Yellow Tail has the highest density of xanthophores, and the XDH level in the yellow varieties is likely to be correlated to the density of xanthophores rather than iridophores. Finally, the XDH expression in male Doublesword, which has a high density of iridophores, was higher than in juvenile and female Doubleswords. In conclusion, caudal fin XDH mRNA levels appear to be higher in guppy varieties that have a higher density of iridophores, with the exception of the Yellow Snakeskin. 

... 

XDH is involved in the biosynthesis of certain pterins and purines. Isoxanthopterin and xanthopterin are colorless pigments present in the xanthophores of medaka (Hama, 1975), while isoxanthopterin is a yellow pigment in isopods (Nakagoshi et al., 1992) and xanthopterin is a yellow pigment in zebrafish (Guyader and Jesuthasan, 2002). The leucopterin and 7,8-dihydroxanthopterin are reflective pigments in gizzard shad and percids (Oliphant and Hudon, 1993). In wild-type axolotl, the inhibition of XDH with allopurinol induced the reduction of xanthopterin, isoxanthopterin, biopterin, and sepiapterin in the xanthophores. Furthermore, they showed a significant reduction in xanthophores and iridophores, while melanogenesis was enhanced (Frost and Bagnara, 1979; Thorsteins!dottir and Frost, 1986). In zebrafish, there is a peridine pathway associated with xanthophore pigments (Ziegler et al., 2000). An alternative XDH pathway, which catalyzes 7-oxobiopterin synthesis from 7,8-dihydrobiopterin and biopterin from the precursor sepiapterin, has also been established. 

The observation that the inhibition of XDH activity enhanced melanin synthesis in conjunction with the reduction of xanthophores and iridophores in axolotl suggests that the down-regulation of pigmentation in these chromatophores is accompanied by the up-regulation of melanogenesis. This linkage was also observed in the zebrafish mutant nacre, in which the gene encoding the microphthalmia transcription factor carries a null mutation. The mutant has a body pattern of 40% more iridophores and an absence of melanophores (Lister et al., 1999). In the guppy, our Northern blot analysis indicated different signal intensities in caudal fins, but the expression levels were too low in some varieties to be meaningfully compared. Guppy varieties that lack iridophores consistently show lower levels of XDH expression. In contrast, it appears that XDH is more highly expressed in guppy varieties that have a high proportion of iridophores. However, this association is not consistent, particularly in the Red and Yellow Snakeskin, and Blue Metallic guppy varieties, in which the XDH expression levels remain low despite the presence of substantive iridophores." (Ben et al. 2003). 

Hudon et al. (2003) studied the carotenoids present in guppies' integument and they found: (1) "Thus, guppy skin mainly contains esters of tunaxanthin alongside a few unesterified monosubstituted carotenoids with e end-rings."; (2) "These results suggest that males deposit a greater fraction of ingested carotenoids in the skin than do females and that males route the surplus carotenoids into the orange spots as opposed to shunting carotenoids away from general pigmentation and into the orange spots."; (3) "Surprisingly, the whole-skin carotenoid extracts of females were more similar in peak wavelength to the male orange spot extracts than to the male nonorange spot extracts."; (4) "The peak wavelength of whole-skin extracts was more similar between the sexes within streams than among streams within a sex, as reflected by a significant correlation between the sexes across streams. This suggests an effect of the local environment."; (5) "This would seem to indicate that beta-carotene has a stronger influence on peak wavelength of skin carotenoids than does zeaxanthin. Similar results were obtained for the nonorange skin of males. In contrast, the peak wavelength of the orange spots of males did not appear to be related to variation in periphyton carotenoid composition." 

"The carotenoid pigments in the orange spots of male guppies were very similar to those in the non-orange skin of both sexes (mostly esters of tunaxanthin), which is surprising given that the orange spots contrast conspicuously with the remainder of the integument and are absent altogether in females. Nevertheless, the carotenoids in the orange spots differed from those in non-orange skin in terms of pigment concentration, peak wavelength, and the dependence of these parameters on the environment. It is probable that the orange spots and outlying skin differ in other ways, for example, properties of the iridophore layer, that contribute to the color difference (G. F. Grether, personal observation). 

The concentration and peak wavelength of carotenoids in the skin of guppies were found to depend, respectively, on the amount and composition of carotenoids in the algal food base (periphyton), although these effects differed somewhat between the sexes and between the orange and non-orange skin of males (Fig. 7). The males in our sample had 5.4–8.9 times more carotenoids in their orange spots per square millimeter of skin surface than in the rest of their integument. This was not accomplished at the expense of the general integument, which was endowed similarly to the general integument of females. Instead, males appear to use more of the carotenoids they ingest than do females and channel the "surplus" carotenoids into the orange spots. In LCA streams, males have relatively low concentrations of carotenoids in their orange spots, but elsewhere in the integument the concentration of carotenoids does not differ between LCA and HCA streams (Grether et al. 1999; this article). This suggests that general pigmentation needs are met before sexual ones even though the amounts of carotenoids found in the orange spots are greater than those in the general pigmentation. Apparently, carotenoids are not used in a strict order related to amounts used but instead according to an adaptive hierarchy. Carotenoid pigments are used for a variety of other purposes and occur in a wide range of tissues, as well (Needham 1974). Other proposed functions of carotenoids, such as immune system enhancement (Lozano 1994; Møller et al. 2000), may rank higher in priority than general pigmentation but require even smaller quantities of carotenoids. 

The carotenoid composition of the non-orange skin of both sexes was correlated with the relative abundance of the known usable carotenoids lutein, beta-carotene, and, to a lesser extent, zeaxanthin in the periphyton (Fig. 7). An association could arise if these carotenoids were deposited unmodified in the integument, or in this case served as precursors for the carotenoids present in the skin. Tunaxanthin is the main carotenoid present in the skin of guppies (and the end product along a carotenoid biochemical pathway in fish), but other carotenoids were present as well. 

Lutein can be expected to contribute to the deposition of tunaxanthin in the skin of guppies because it is a direct precursor of that carotenoid (Miki et al. 1985). The contribution of beta-carotene is harder to explain, as it is many steps removed from tunaxanthin and not a known precursor of the yellow carotenoid. However, the identification of two probable metabolites of beta-carotene on the path to tunaxanthin in the skin of the fish examined (compounds B and C) suggests that such conversion may be taking place. Compounds B and C may appear in the skin because of their relatively high hydrophobicity. 

Beta-carotene, and probably other dietary carotenoids, also yields pigments absorbing at longer wavelengths than tunaxanthin, judging from the peak wavelength of skin carotenoids above 438 nm (tunaxanthin’s peak wavelength) in many fish. These were apparent particularly in streams with higher relative abundance of beta-carotene. Dietary carotenoids may be expected to contribute to different degrees to skin pigment composition as a function of the ability of guppies to absorb, transport, and process these carotenoids. 

The slope of the relationship between skin and dietary carotenoids suggests that a slight increase in the peak wavelength of ingested carotenoids would result in a disproportionately greater increase in the peak wavelength of skin carotenoids (note the different scale of the horizontal and vertical axes in Fig. 7). This observation would seem to indicate that the pigment composition of the general integument, especially in females, is very sensitive to variation in the relative abundance of lutein and beta-carotene in the diet. This sensitivity was further borne out by the highly significant correlation between peak wavelength of skin carotenoids and the beta-carotene/lutein ratio of the periphyton. The latter result could indicate that beta-carotene or a metabolite counterbalances lutein in some of the processes involved, although other explanations are possible. Whether this norm of reaction as a function of pigment availability in the diet is adaptive or not will require further investigation. 

Another surprising result of the present study was that the pigmentation of the non-orange spot fraction of male guppies is not equivalent to that of the general integument of females, considering that the two may play similar roles, such as crypsis or protection from UV light, for example. Instead the pigment composition of the integument of females was consistently closer to that of the orange spot fraction of males than to that of the skin outside of the orange spots. At a physiological level, the difference in pigment composition of the general integument of males and females could arise from the processes responsible for the differentiation in pigment composition between the two color fractions of the males if, for example, the two skin areas of males differed in their selectivity of carotenoid uptake, the orange spots preferentially absorbing longer  wavelength-absorbing carotenoids, leaving shorter wavelength absorbing pigments for the rest of the integument or vice versa. Functionally, the difference in pigment composition of males may also provide for added visual contrast between the orange spots and the surrounding skin. reasons, besides phylogenetic constraints, for the use of tunaxanthin. 

Although carotenoid composition of the non-orange spot fraction of males and the general integument of females was correlated with diet, that of the orange spot fraction of males was not. This suggests that carotenoid composition in the orange spot of males is controlled by different factors than those operating in the general integument." Hudon et al. (2003).

Right now, and as far as I know, there are three non genetic ways we can boost guppies colors: (1) to work with their adaptation to background colors and lightning, I'll return to this point later; (2) use chemicals like hormones or other stuff that causes pigment aggregation or dispersion, this is, at least, taken as a faulty posture in shows...; and (3) supplement their diet with plenty carotenoids. See Ako et al (no date).

When I saw this report that guppies skin contain tunaxanthin I figured about the use of tuna flesh in pastes. Fortunately sounds that guppies can produce tunaxanthin from lutein and beta-carotene. Fox (1976), p 398, cites that Artemia convert dietary beta-carotene to echinenone and, finally, to canthaxanthin and that these are the only carotenoids present in their eggs and nauplii. At pages 402-403 he say that Crozier (1997) was the first to show the ability of an animal to convert tunaxanthin  to astaxanthin, either straightly or via zeaxanthin; and that "Crozier's findings may be summarized as follows, showing a range from red down to olive drab, through which astaxanthin esters decreased corresponding with increases of neutral xanthophylls, principally tunaxanthin." Well, how could we describe the color of the general integument of a wild guppy, olive drab??? Isn't possible that breeders of red guppies selected fishes able to produce astaxanthin instead tunaxanthin??? Should I figure about the a possible inversed pathway from astaxanthin to tunaxanthin since most guppy breeders swear that feeding BBS improve colors of any strain??? See NRC (1993). What about e(i)chinenone??? See Aquamedia (no date). Food for thought...          

On mutations effecting xantho-erythrophores we have the so called blaus. (1) European blau, is a recessive gene which prevent the deposition of red and yellow pigments on the body and of, at least, yellow pigments in the caudal fin; (2) Asian blau is a co-dominant gene which in "single dose", i. e. heterozygose, prevents the deposition of red pigments on the body and caudal (see micrographies of red and blue grass taken Mr. Kobayashi 's collaborator at guppy.to site) and while in "double dose" prevent the deposition of both red and yellow pigments on the body and caudal; and (3) Hell blau, a rare recessive gene which prevents the deposition of red and yellow pigments on the body and, as far I know of red pigments in the caudal fin.

Leucophores.

These cells render white color to guppies integument because they possibly enclose purine compounds in  leucossomes. "Purine compounds are derived in part from nucleoproteins of the diet, although some animals are undoubtedly able to synthesize them, probably from amino-acids arginine and histidine.".  "In strict sense these compounds, being colorless, are hardly to be classed as true biochromes. However, since certain member of the series, notably guanine and uric acid, contribute opaque whiteness, glistering silvery aspects or even iridescent colours to animals in the lower groups, a brief consideration of them probably belongs in this treatise." Fox (1976), p 289.

"Of the purines which constitute the white or silvery "pigment" in the integumentary structures, guanine is the greatly predominating member, while uric acid is observed in some instances. ... Leucophores are guanophores in which the guanine is deposited in minute granules which are motile, and thus migratory within the cell."  Fox (1976), p 289-290. 

Takeuchi (1976) was the first one to describe guppies' leucophores: "With dark-field microscopy two distinct types of reflecting chromatophores were observed to be numerously distributed in the guppy tailfin. (Note: They analyzed  guppies having whitish tailfins obtained from commercial hatcheries). ... the leucophores are larger (Note: Authors were comparing size of leucophores with the size of iridophores), highly dendritic, and exhibited a dull whitish color. The leucophores gradually contracted 20 or 30 min after tailfin was cut off. ... In the tail fin taken from fish reared in melanine-saturated water for two days, the leucophores had completely disappeared. 

...  

The dendritic leucophores are very large, having large and convoluted nuclei and cytoplasms with a large number of characteristic pigment granules. Nuclei usually lie at periphery of the cell. Granular and fibrous elements are distributed throughout the nucleoplasm, with a few aggregates of granular elements suggestive of nucleoli or chromatin.

The spherical or ellipsoidal pigment granules of the leucophores, measuring about 0.5-0.8 mm in diameter, are enclosed by a double membrane  with a rather broad intervening space; the outer membrane strongly resembles the delineating membranes of other cytoplasmatic organelles such as mitochondria and endoplasmic reticulum, but the inner is reminiscent of lamellae, appearing to originate from the fibrous materials so numerous in the internal parts of these granules. 

...

For the identification of leucophores, the guppies reared for one day in melanine-saturated water were sacrificed and pigment granules of each chromatophores were examined. Only the pigment granules in the chromatophores described above as leucophores exhibited conspicuous degenerative changes. Outer limiting membranes of the granules were torn off and frequently disappeared. Inner lamellae and internal fibrous materials were coagulated into small, thick membranous materials." Takeuchi (1976).

Takeuchi (1976) did not reach a conclusion about the nature of the pigments found in guppies' leucossomes but I don't think that it is guanine as stated as most possible by Fox (1976), despite later author was not specifically making mention to guppies. (1) If guppies' leucossomes were filed with the same material found in their iridophores I would expect that, at least partially, both presented the same response to melanin treatment; and (2) In Medaka, that is a classical species in chromatophore's study (Chromatopapers 1 and Chromatopapers 2), so much more studied than guppies; that carry some similarities with guppies in the carotenoid content of xanto-erythrophores (Hirao et al., 1969) and in the occurrence of similar mutations effecting color cells (Takeuchi, 1975b); was found that leucossomes are filled with uric acid (Hama, 1975).  

Iridophores.

Iridophores and leucophores are the cells which produce the so called structural colors. The cells of both above mentioned chromatophores enclose pigments of similar origin (purines) and accordingly some sources the also enclose the same "pigmentary compound". "When the purine material is in fixed crystalline platelets, manifesting a scintillating iridescent or metallic appearance, the cells containing them are termed iridophores or iridocytes." Fox (1976), p 290.

Takeuchi (1976) was the first one to microscopically describe guppies' iridophores too: "The iridophores are smaller (Note: When compared to leucophores), having a metallic iridescent color, ... The cytoplasm of iridophores of the guppy contain a large number of laminated cisternae or "reflecting platelets". Each of these platelets is surrounded by its own limiting membrane. Their lengths vary from 0.7 to 0.2  mm and their diameters are approximately 35 mm. The platelets are found in several groups. Within each group several platelets, sometimes up to 17 or 18, are stacked up with an interval of about 150 mm. Angles among groups are variable. Many of the platelets are split longitudinally, exhibiting empty spaces. Electron microscopy reveals that these longitudinal splittings are initially narrow, but the electron bean rapidly enlarged them in various degrees, markedly affecting their appearance. Therefore, carbon-coated sections were used for detailed observation of these platelets. The longitudinal splitting extended either the entire length of the platelets, or was interrupted at irregular intervals. In the center of platelets that were not split, very thin lamellas materials extended over the entire axis. In some platelets in where only partial longitudinal splitting had occurred, the interconnection between these lamellae and the splitting was clearly noticed."  Takeuchi (1976). 

Recently Gundersen and Rivera (2005) also studied this subject but I was not yet able to get a copy of their article. 

Fox (1976), chapters IV and V, provide us an excellent review on the physical mechanisms by which structural colors are produced, i. e. scattering, diffraction and interference.

Scattering, sometimes called Tyndall scattering, occurs when a light bean impinges upon particles with a mean diameter less than about 0.7 m and that are not orderly oriented, actually they are in a state of random dispersal . Diffraction is also produced by particles which are not orderly oriented but in this case the particles are larger, exceeding about 1 m (Fox,1976,p 21, 41). "The kind and intensity of colour, and the intensity of Tyndall cone, observed when a bean of light is passed through a light-scattering medium, depend upon the degree of fineness of the small particles in the suspension, and upon the magnitude of difference between the refractive index of the particles and that of the fluid medium. Sufficiently small particles yield violet colours of striking intensity and purity; increasingly larger ones give a series of deep to pale blue colours, passing to white when approaching 1 m in the mean diameter." 

"Mason (1923a) has emphasized the importance of actually seeing and devoting some optical study to Tyndall blues in order to understand the principles by which underlie their manifestation, and to  differentiate them clearly form the other blues aspects. A summary of the characteristics of Tyndall blues and the properties of media which lead to their manifestation includes the various points listed by Mason:

(1) The manifestation of blue color depends upon the presence of minute bodies or other optical heterogeneities whose refractive index differs from that of surrounding medium.

(2) The diameter of light-scattering bodies is of the order of the wave-length of blue light, i. e. approximately 0.6 m or less.  

(3) The scattered light is blue, the transmitted fraction yellowish or reddish; consequently the blue light is visible only by reflexion. The object appears reddish, brown or grey by transmitted light.  

(4) Relative sizes of particles determine depths or shades of blue; very small micelles or air-spaces yield violet hues, the larger ones whitish blue or white.

(5) The scattered blue light is more or less polarized in a plane normal to the direction of the incident beam; the degree of polarization depends upon particle size.

(6) The intensity of scattered light is an inverse function of the fourth power of the wave-length.

(7) No iridescence or conspicuous changes of colour occur over wide variations in the angle of reflexion. (Distinction from interference-colours).

(8) No blue material is extractable or precipitable. (In contrast with pigments.)

(9) The colour may be destroyed by mechanical means, such as grinding or pounding. (In contrast with pigments.)

...

(11) The blue colour may be abolished by mechanical or chemical removal of the underlying melanin layer, but can be restored by the application of any suitable dark material to the under-surface as a substitute light-absorbing screen." Fox (1976), p 23-24.

"Brilliant blue colours abound in many species of fish, and these striking manifestations are founded upon the fact that melanophores, packed with the dark pigment melanin, are present in the integument, and that surrounding, intermingled, and specially overlying these pigment cells are other cells filled with minute crystals or other aggregates of purine material, notably guanine. The contents of these guanocytes (also called guanophores, iridocytes, iridosomes or leucophores) are of such minute size, and of refractive index sufficiently different from that of the fluid cell contents, as to render possible their scattering and reflexion of blue light, while the transmitted light fraction is absorbed y the melanin beneath (but see footnote, p. 27)". This is the footnote at page 27:"The many brilliant specular blue or changeable blue-green patches exhibited by fishes' skin are doubtless interference colours (see Chapter V)" Fox (1976), p 25-27.

"Numerous fishes show black areas due to melanin in the upper layers of the skin, blue regions arising from the phenomenon just described or resultant green when yellow carotenoid bearing cells (xanthophores) are superposed on the blue loci. Furthermore, such a fish may exhibit the brilliant changeable colours of iridescence when the angle of the observer's vision is altered. The later aspect due to interference of light by the surfaces of the thin scales themselves, or by these and the thin layers of guanine particles not so placed or so oriented as to manifest the blue of scattering." Fox (1976), p 26.  

Interference occurs when a light beam impinges upon particles with a mean diameter less than about 0.7 m and that are orderly oriented. "Table I indicates the fact that the iridescent colours, like the Tyndall blues, occur preponderantly in the integument, in certain of its derived structures and in internal membranes. Both types of colour manifestation depend upon the disintegration of light by structures whose unit dimensions are necessarily of the order of the wave-length of visible light, i. e. 0.7 m or less. But we have seen that the manifestation of scattered blues light arises from optical heterogeneities in a state of random dispersal, whereas iridescent colours result from a condition which might be said to be in direct contrast to the latter, in that the basic chromogenic units are thin films or laminae, arranged in uniform layers by periodic secretion and deposition. The blue colours of scattering are delicate and easily masked by reflected white light, so that a substratum of some absorbing material, such as melanin or haemoglobin, is necessary for their manifestation. While melanin often likewise underlies multiple thin transparent laminae, thus similarly absorbing the transmitted light and so intensifying the iridescence of the reflected fraction, the pigment is not always present; nor does this fact preclude the manifestation of interference colours." Fox (1976), p 41.  

In order to produce green and purple colors, when Tyndall scattering is the origin of the blue structural color, a organism must deposit yellow or red pigments, respectively, above the iridophores. On the other hand, if the structural color arises due interference there is a possibility that green and purple be created changing the thickness and distance between plates (see Table 1 and Walin (2002), figure 9).

Table 1 - Schematic relation of wave-length and colour to laminar thickness, interlamellar air-space and angle of incidence; pile of twelve plates, each of refractive index 1.5, after Anderson and Richards. Adapted from Fox (1976), p 57.

Transparent scale and China glass are mutant guppies which do not deposit guanine platelets in peritoneal wall. 

Releasing of this issue is too close to get deeper, so in order you can understand how chromatophores work together to produce colors I strongly suggest you read Bagnara et al (1968) and Grether et al. (2004).  

References.

Anderson, T. F. and A. G. Richards Jr. 1942. An electron-microscope study of some structural colours of insects. J. Appl, Phys. 13:748-758.

Crozier, G. F. 1967. Carotenoids of seven species of Sebastodes. Comp Biochem. Physiol. 23:179-184.

Fox, D. L. 1976. Animal biochromes and structural colors. University California Press, Berkley.

Goodrich, H. B., N. D. Josephson, J. P. Trinkaus and J. M. Slate. 1944. The cellular expression and genetic of two new genes in Lebistes reticulatus. Genetics, 29:584-592.

Hurst, T. D. 1980. An introduction to the chemistry and biochemistry of pyrimidines, purines and pteridines. John Wiley. New York.

Mason, C. W. 1923, Structural colors in feathers. I. J. Phys. Chem. 27:201-251.

Nakajima, M., E. Shinohara and Y. Fujio.1999. Fluorescent chromatophore detected in the guppy Poecilia reticulata. Zoological Sci.,16:745-747.

Nayudu, P. L. and C. R. Hunter. 1978. Genetic control of melanophore ultrastructure in Poecilia reticulata. J Fish Biol 13:453-456.

Nayudu, P. L. and C. R. Hunter. 1979. Cytological aspects and differential response to melatonin of melanophore based color mutants in the guppy, Poecilia reticulata. Copeia 1979(2):232-242.

Sumner, F. B. and N. A. Wells. 1933. The effects of optic stimuli upon the formation and destruction of melanin in fishes. J. Exp. Zool. 64:377-403.

Takeuchi, I. K.1975. Pterinosomes in erythrophores of the guppy, Lebistes reticulatus. Jap. Ichthyol. 22:43-45.

Takeuchi, I. K.1976. Electron microscopy of two types of reflecting chromatophores (iridophores and leucophores) in the guppy, Lebistes reticulatus Peters. Cell Tiss. Res 173:17-27.

Ueshima, G.,  M. Nakajima and Y. Fujio. 1998. A study on the inheritance of body color and chromatophores in the guppy, Poecilia reticulata. u J. Agric. Res. 48(3-4):111-122.

Yamamoto, T. O. 1975. The medaka, Oryzias latipes, and the guppy, Lebistes reticularis. In: King, R. C. (ed) Handbook of Genetics Vol 4. Vertebrates of genetic interest. Plenum Press, New York, USA. pp 133-149.