Females of many animal species show mate preference based on attributes like bright coloration, big antlers, or long tails. However, guppies—the small, freshwater fish often seen in home aquariums—are quite unique in that within a single species, different populations and even different individuals can show distinct mate preferences. For this reason, guppies have become a classic model for the evolution of female mate choice.
The wild Trinidadian guppies (Poecilia reticulata)
The wild guppies (Poecilia reticulata) studied by Benjamin Sandkam (Simon Fraser University, British Columbia, Canada) are native to the mountain streams of Trinidad, and have sexually dimorphic coloration—the females are uncolored, while the males show color variation—especially in the presence and size of red-orange markings on the males (Figure 1).
These guppies are only a couple of centimeters long and occur in small streams high in the Trinidadian mountains that are only about 1/3 m deep and 1 m wide. As the streams descend the mountains, they become larger and pass over a series of waterfalls. Below the falls, the water is deeper, and there are larger pools that provide habitat to larger and more numerous predators. It is these areas below the falls that are considered high-predation environments for the guppies, while the areas above the waterfalls are considered low-predation areas, with the falls preventing predators from moving upstream into these areas.
The male guppies show strong color variation across different populations of the same species, with strongly red-orange colored males (Figure 1A) more prevalent in low-predation areas, and less colorful males (Figure 1B) in high-predation areas.
The females prefer to mate with the more colorful, red-orange males, but, interestingly, the extent to which this is true differs across guppy populations within this species, in this region (Figure 2). Generally, females in low-predation sites more strongly prefer red-orange males than those females in high-predation sites. These observations led to questions about the genetic factors that drive evolution of divergent mate selection behavior in guppy populations.
The relationship between color vision and mate choice
Questions about genetic involvement in mate selection intrigued Benjamin Sandkam, a graduate student in Dr Felix Breden’s laboratory (Simon Fraser University, British Columbia, Canada). Ben is building his research career on explaining the evolution of the mechanisms underlying behaviors. He has a particular interest in how mate preferences are passed on across generations. Given that many species do not teach these preferences to their offspring suggests that there is an underlying genetic force that accounts for inheritance of mate preference.
For coloration to play a role in mating decisions, organisms must have the ability to detect and discriminate different colors, commonly called “color vision”. Color vision is accomplished by comparing signals from different cone cells in the retina, which vary in the wavelength of light to which they are most sensitive . The most variable component involved in tuning cone cells is the transmembrane protein called an opsin. Guppies possess 9 different opsins, with 4 separate Long Wavelength-Sensitive (LWS) opsins responsible for red-orange color detection.
Sandkam and Breden wanted to examine whether there was a causal relationship between visual tuning and mate choice at the population level within Poecilia reticulata. To do this, they collected guppies from low- and high-predation populations in 2 distinct watersheds in the Northern Range Mountains of Trinidad. Colonized independently from low- to high-predation populations, these watersheds provide a prime example of parallel evolution. Female guppies consistently show differences in mate choice for red-orange colored males between these low- and high-predation populations.
Specimen collection took into account time of day and light intensity. The scientists monitored other environmental parameters, including dissolved oxygen concentration, total dissolved solids (water clarity), as well as the water temperature, conductivity, pH, and salinity. Specimens were dissected in the field, and eye tissue samples preserved in RNAlater® Reagent (Thermo Fisher Scientific) and returned to the laboratory for processing. Details regarding experimental design, methods, and analysis can be found in Sandkam, Young, Breden (2015) .
Opsin expression levels illuminate differences in color vision
The researchers tested whether variation in the ocular sensory system exhibited parallel variation to mate preferences by assessing the differences in allele frequency and expression of LWS opsins. Sandkam used qPCR to generate the expression data for this study. PrimeTime qPCR Assays (IDT) were designed for 9 opsin genes (LWS-1, LWS-2, LWS-3, LWS-R, RH2-1, RH2-2, SWS1, SWS2A, SWS2B), 1 rhodopsin gene (RH-1), and 3 housekeeping genes (β-actin, cytochrome C oxidase subunit 1, and myosin heavy chain), with a primer or probe positioned to span intron-exon boundaries wherever possible. (Interestingly, some of the guppy opsin genes contain no introns, so a thorough DNase treatment of the extracted RNA samples prior to reverse transcription and qPCR was imperative.)
5′-FAM–labeled probes contained the ZEN™ Internal Quencher as well as a 3′–Iowa Black® Quencher. Use of double-quenched probes can decrease fluorescence background and increase signal. Sandkam notes, “Some of the opsin gene sequences, and thus qPCR assays, were very similar. But the PrimeTime qPCR Assays, used in conjunction with the ZEN/Iowa Black Double-Quenched Probes, were really helpful, because they were so specific that I could design locus specific assays based on very small sequence differences across these different loci.” (See Product focus—Double-Quenched Probes, right, for more about this internal quencher.)
Sandkam et al. (2015)  designed 3 gBlocks® Gene Fragments (IDT) to determine the PCR efficiency of each assay. The gBlocks Gene Fragments each contained the sequences for 3–4 of the genes being assayed, from 10 bp upstream of the forward primer to 10 bp downstream of the reverse primer. The 3 gBlocks Gene Fragments were mixed in equal proportions and brought to 0.001 ng/µL, generating a control with equal ratios of all the opsin and housekeeping genes. “Using gBlocks Gene Fragments to create my qPCR standards saved me months of time, probably most of a year, not to mention the money saved by not having to construct and clone them,” said Sandkam. (See Product Focus—gBlocks Gene Fragments (right), for more about these double-stranded synthetic DNA fragments.)
Thirteen 10 µL assays were run in triplicate for each of the 288 guppy specimens, using Brilliant III Ultra-Fast qPCR Master Mix (Agilent Technologies) on an Applied Biosystems 7900HT qPCR machine (Thermo Fisher), which meant there were 11,232 experimental sample reactions plus 416 reactions of negative controls, or 11,648 reactions total, run as 32 separate 384-well plates.
Differences in color vision were assessed by measuring the proportion of total opsin expression, taking into account assay efficiency and housekeeping gene expression (Figure 3). Proportional opsin expression provides a measure of color vision, as opsins are the major differentiating character of cone cell types, and color vision is accomplished by comparing the signal from different cone cell types [2–4]. Opsin expression relative to the housekeeping genes showed how the opsins were differentially regulated. Of the various environmental parameters (see above) measured by the scientists, none corresponded with the expression differences of LWS seen in the test guppy populations .
Validation of the sensory exploitation model of population divergence
The researchers used MANOVA and ANOVA statistical analyses of the qPCR data to confirm that color vision varies across guppy populations and that low-predation guppies develop greater color vision for detection of red-orange coloration. Specimens collected from independently colonized watersheds showed that guppies expressed higher levels of both LWS-1 and LWS-3 (the most abundant LWS opsins) in low-predation compared to high-predation populations at a time that corresponds to differences in cone cell abundance. Through additional experiments, the researchers observed that the frequency of a coding polymorphism also differed between low- and high-predation populations. Details of the statistical analysis and allele frequency experiments can be found in Sandkam, Young, Breden (2015) .
Together, the researchers’ results support the Sensory Exploitation model for the evolution of female mate preferences; that is, the variation in peripheral visual systems, rather than changes to neural processing in the brain, can drive mate preference variation. The work also proposes important candidate genes involved in the genetic basis of female preference variation. This work provides one of the best examples so far of mate preferences within a species covarying with differences in color vision.