Journal of Bioeconomics

, Volume 20, Issue 1, pp 125–140 | Cite as

Experimental evolution of color preference for oviposition in Drosophila melanogaster

  • Mellissa Marcus
  • Terence C. Burnham
  • David W. Stephens
  • Aimee S. Dunlap
Article

Abstract

Preferences are the foundation of economics. Preferences are taken by economists as fixed by some implicitly biological process. In recent decades, behavioral economics has documented the divergence between the nature of human preferences and the assumptions of standard economics. In this study, we use the tool of experimental evolution to study the evolution of color preferences in fruit flies (Drosophila melanogaster). In particular, we select for a preference for laying eggs on the color aqua. We find that the flies evolve to lay more than twice as many eggs on aqua. However, this evolution occurs entirely because the flies lay more eggs overall. The flies in this study, do not evolve to lay a higher percentage of eggs on the selected color, aqua.

Keywords

Adaptation Preference theory Experimental evolution Evolution Behavioral economics selection 

1 Introduction

This is an experimental investigation into the evolution of preferences. The long-term goal of this research program is to improve economics by introducing experimental evolution as a tool for empirical studies of preferences (Burnham et al. 2015). This paper reports on a laboratory study of the evolution of preferences over color in fruit flies of the species Drosophila melanogaster (D. melanogaster hereafter).

Preferences are at the foundation of economics and yet are exogenous to the field—taken as fixed by some unstated, implicitly biological, process (Burnham et al. 2015). Tastes are “the unchallengeable axioms” of economics (in latin, De gustibus non est disputandum), and research on the origin of preferences is left to, “whoever studies and explains tastes (psychologists? phrenologists? sociobiologists?)” (Stigler and Becker 1977, p. 76).

Leaving the foundation of economics to non-economists might be pragmatic if the nature of preferences were themselves not disputable. However, within economics exactly the opposite is true. The axioms of economics are a matter of sharp disagreement between neoclassical and behavioral economists (Tversky and Kahneman 1974; Kahneman and Tversky 1979; Kahneman et al. 1991).

While we are not aware of any previous empirical studies on the evolution of economic preferences, there has been some evolutionary modeling of preferences (Rubin and Paul 1979; Hansson and Stuart 1990; Robson 1994; Rogers 1994; Brennan and Lo 2011; Robson et al. 2012; Collins et al. 2016). There is also a considerable theoretic literature on cultural evolution related to economic behavior (Panchanathan and Boyd 2003). Finally, there is a substantial experimental evolution literature that was not designed with economics in mind, but has important implications for the field (Burnham et al. 2015).

Experimental evolution provides a novel tool for the empirical study of preferences. Controversial issues within economics, that have not been resolved of the last half century, can be studied, albeit with organisms that are quite distant phylogenetically from humans (Burnham et al. 2015).

Some of the important issues that can be empirically examined include: Can an economic good be created? Is it possible to experimentally evolve a taste (or distaste) for some aspect of the environment? What are the sources and extent of variation in preferences between individuals? What is the time scale of genetic adaptation to a changing environment? Why do people sometimes have very strong desires for products that are bad for them?

This study uses fruit flies (D. melanogaster) to study the evolution of preferences. Given that fruit flies and humans diverged from each other hundreds of millions of years ago (Peterson and Eernisse 2016), it is reasonable to ask what aspects of human economic behavior can be studied using a fruit fly model.

Notwithstanding the enormous differences between humans and fruit flies, we believe that valuable inferences can be drawn. The aspects of human decision making that utilize conscious calculation may have no similar process in drosophila. However, the fundamental drivers of pleasure and displeasure, the “goods” themselves in economics, were created by the same process in these two disparate species.

Preferences evolved to reflect reproductive payoffs for ancestral organisms. For example, humans derive pleasure from eating maple syrup, and not maple bark, because human ancestors that consumed sugar in maple syrup had higher reproductive success than competing organisms that ate bark. Conversely, termites prefer maple bark to maple syrup because their ancestors faced the opposite reproductive payoffs.

The idea that preferences reflect fitness payoffs dates back at least to Adam Smith and, more recently (and explicitly) to Gary Becker (Burnham et al. 2015). The goal of this study is to investigate how organisms evolve preferences.

With this motivation, we performed an experimental evolution study of color preferences in fruit flies. To select for these preferences, flies exist in an engineered, colorful world where pregnant fruit flies lay their eggs on surfaces that vary in their location relative to particular colors.

The selection for color preferences is implemented by placing different colored discs below a surface where the flies lay their eggs. In the test treatments of this experiment, only eggs laid above an aqua-colored substrate are used to found the next generation. The experiment creates selection for laying eggs on aqua. Thus, any genetic variation that leads to more eggs laid on aqua will be favored. Conversely, any genetic variation that leads to fewer eggs laid on aqua will be selected against.

The rest of this paper is structured as follows. First the methods, followed by results, and ending with a discussion.

2 Methods

Drosophila melanogaster are used in this study of the evolution of color preferences. The selection for color preference is implemented by harvesting eggs, in the test conditions, that are laid over a particular color. D. melanogaster is an extremely well-studied organism, utilized in many evolutionary and genetic studies, and with a well-documented color vision system (Harzsch et al. 2007; Erclik et al. 2009; Gonzalez-Bellido et al. 2011).

As noted, the selection is based on oviposition by females. Understanding how females decide where to lay eggs has been an area of growing interest, both evolutionarily and neurobiologically (Yang et al. 2008; Joseph et al. 2009; Miller et al. 2011; Abed-Vieillard et al. 2013; Dweck et al. 2013). Work on the role of color in oviposition preference is more sparse (Carfagna and Lancieri 1971; Solar et al. 1974; Solar and Ruiz 1979), however enhanced color learning has been shown to evolve under certain conditions (Dunlap and Stephens 2014).

The colors utilized in this experiment were chosen because previous studies determined two optima peaks for D. melanogaster wavelength discrimination at 420 and 495 nm (Salomon and Spatz 1983; Heisenberg and Wolf 2013) and peak absorptions in the UV, at 475 and 515 nm. (Salcedo et al. 1999; Washington 2010). We verified that the flies are able to recognize, and discriminate between, these wavelengths in our own pilot studies.

The colors utilized are based on the physiology of fly vision, choosing two peak wavelengths of their vision and then a color right in the middle (the aqua). This choice reflects the sensory system of the animals, rather than the environment, since the flies can only make choices about what they perceive.

When it comes to fly vision and oviposition, there is little evidence that Drosophila care very much about color in the wild. It is most likely that olfactory cues override color in most contests and across evolutionary time (Dunlap and Stephens 2014). However, Drosophila have demonstrated an ability to identify colors and make color-based choices in learning experiments in both oviposition (Dunlap and Stephens 2014) and in other contexts (e.g., Brembs and Ibarra 2006; Schnaitmann et al. 2013). The flies care enough about color for us to expect evolution of color preferences, but not so much as for us expect that their preferences are immutable.

We utilized a well-established methodology for associated areas of agar with particular colors. In this method, colored paper disks are placed under petri dishes of clear agar (similar to Dunlap and Stephens 2014). To create the color stimuli, we converted the desired wavelengths to Red–Green–Blue (RGB) hexadecimal input and printed the wavelengths, now hues, using a color laser-printer. These disks were then laminated, and each disk was placed under a 4 cm petri dish, which was then filled with a small layer (1.5 mL) of a clear agar (10 g agar/L, X g sucrose/L).

We designed this experiment to study the experimental evolution of color preferences. In particular, the study creates selective pressure to lay more eggs on aqua. There are two complementary routes the flies could in response to this designed selection for aqua. First, the flies could evolve to lay a higher percentage of their eggs on aqua. Second, the flies could evolve to lay more eggs. Of course, the flies could take both of these routes

2.1 Treatments

We created three test treatments and two control treatments (see Table 1; Fig. 1). In each of the test treatments, flies were presented with stimulus agars of different colors, and in each case the “best” choice was to lay eggs on aqua.
Table 1

summary of three test and two control treatments

Treatment

Green

Aqua

Blue

Selection process for starting next generation

Test 1

Yes

Yes

 

Select exclusively from aqua

Test 2

 

Yes

Yes

Select exclusively from aqua

Test 3

Yes

Yes

Yes

Select exclusively from aqua

Control 1

Yes

Yes

Yes

Select equally from all colors

Control 1

No

No

No

Select equally from all regions

One test treatment included aqua plus green (lower in peak wavelength), another test treatment included aqua plus blue (higher in peak wavelength), and the third test treatment included all three colors—aqua, blue, and green (Fig. 1; Table 1). The first control treatment used color, but equal amount of eggs were chosen from each petri dish of agar presented to flies, regardless of associated color. The second control treatment did not use any color and, as in the first control, equal amount of eggs were chosen from each dish of food.
Fig. 1

The panels give the selection treatments. In treatments a and b, flies were presented with two colors, with all eggs being collected from aqua in each case. In these cases we imagine that flies are only able to experience part of the full fitness landscape. The dotted line depicts our predictions for these treatments: if evolved fly preference is transitive, flies should show an enhanced preference in the direction of increasing fitness. This would be analogous to a supernormal stimulus response. Treatment c presents flies with the full fitness landscape; here choosing aqua is again the best choice with the highest fitness, but it represents a moderate level of stimulus (choosing hues above and below aqua results in no fitness). Treatment d is a control for choice: egg choice on every color results in equal fitness

2.2 Fly husbandry and starting populations

Each of the treatments and each of the controls had 12 replicates, for a total of 60 different evolving population lines. The starting population was originally wild caught in Michigan (from Ian Dworkin), and has been maintained in large populations since capture. In beginning of the experiment, we collected 360 vials of 80 eggs each and randomly allocated them among the 60 lines, starting new populations. The populations were maintained under selection for 28 generations.

For every generation, we reared flies at \(24\,^{\circ }\hbox {C}\) for 10  days after spending 12 h at \(14\,^{\circ }\hbox {C}\) to delay development long enough to allow for extra time in photographing and moving eggs. Upon emergence as adults, we moved each population into a clear plastic cage (30 cm long \(\times \) 16 cm wide \(\times \) 10.5 cm high) containing two petri dishes (100 mm) with 50 mL of standard cornmeal–molasses fly food. The cages were placed into a walk-in climate chamber (\(24\,^{\circ }\hbox {C}\)), under banks of cool white LED lights (Hitlights©) for 3 days, undisturbed. All selections and testing occurred when flies were 4 days post-emergence as adults, when fecundity is at its peak.

On the fourth day post-emergence as adults, we presented flies with the stimulus disks and plates and allowed them to make choices of where to lay eggs. The dishes were presented at the bottom of the cage, through a sliding floor that replaced the original two large petri dishes of food. There were six possible locations for the stimulus dishes. Two dishes of each color were placed into the cage. For the two-choices treatments, this meant four dishes, and for the three-choice treatments, six dishes.

All petri dishes were labeled with unique serial numbers on the side with red permanent ink; this kept the colored stimuli from being obstructed in any way. The locations of the plates were randomized, with the rule that two identical colored plates could not be next to each other, and these assignments were balanced for each population throughout the experiment. All locations, plate numbers, and color assignments were recorded.

The females of each population were then given time to make their choices as to where to lay eggs. For the first five generations this time was 1.5 h, but was modified in all remaining generations to 3 h, due to the populations laying fewer than the needed 480 eggs to perpetuate the next generations during those first five generations. After the 3 h had passed, the petri dishes were removed from the cages.

Following photography, we completed the selection by collecting eggs from the plates corresponding to the treatment for each population. The eggs were collected along with the thin layer of agar, and placed into vials containing standard cornmeal–molasses food. We collected 480 eggs for each population, split across six vials (80 eggs each). For control populations not experiencing color stimuli, we collected eggs from regular food plates that had been placed into the cages during the selection period, and reared them identically to treatment populations. Each generation was reared at \(24\,^{\circ }\hbox {C}\) for 10 days after spending 12 h at \(14\,^{\circ }\hbox {C}\). Emerging flies were then placed into the clear cages, and the selection began anew.

2.3 Post selection assays

Flies in each of the treatments were selected in different choice contexts, thus to compare them equally we tested naïve subsets of each population in a series of assays. We tested each population in assays mimicking the various choice scenarios of the selections: (1) green and aqua, (2) blue and aqua, and (3) green, aqua, and blue. Locations were randomized as during the selections, with two petri dishes per color. We designed a fourth assay to test the preferences of flies in a larger set of presented colors, to include a color at both extreme ends of the distribution, which flies did not experience during selections. In this assay we presented flies with yellow, green, aqua, blue, and violet, with a single petri dish for each, and locations randomized. We tested three replicates of each population for each of the four assays.

Because of the logistics of completing the large numbers of individual assays, we tested populations after differing number of selections. We tested flies following 21–24 generations of selections. To make accurate comparisons between treatments, we tested equal subsets of replicate populations from each treatment (3 per treatment) following each of the four generations until all 12 replicate populations for each treatment had been tested. In sum, each of the 12 replicate populations for the four treatments and control, were tested three times each in each of the four assay types.

To collect the flies for these assays, we first presented each population with a fresh food plate for 12 h immediately following their usual selection of agar plates and color choices. We then collected eggs directly off the food, to fill 18 vials with 80 eggs each (enough for six vials for each of the three assay replicates). Flies were reared identically to the selection conditions, moved into cages upon emergence, and tested at an equivalent age as flies in selections (4 days post-emergence). Color choices were presented as described above, with plate numbers recorded along with the color, cage replicate, assay, and population. Flies were given 1.5 h to make their choices on where to lay eggs. No eggs were collected following these assays. All plates were photographed following the same procedures as the selections.

2.4 Egg counts

After the egg laying window ended, the colored disks were removed and the plates were photographed using illumination from both above and below to obtain a clear image of the eggs. Every plate was counted by the same individual, who is very experienced in these techniques. The treatments and color assignments were blinded to the researcher performing the counts due to the serial number system.

3 Results

3.1 Choices during selections

We first look at the choices of each population within each treatment across the generations of the experiment in terms of choosing to lay eggs on the aqua substrate, using the preference index described in the methods (Fig. 2; Table 2). We analyzed each treatment separately as an ANOVA, looking at the effect of generation, with line blocked as a random factor. We predicted a positive effect of generation upon aqua preference in the three main treatments. For the control for color, we predict movement towards choosing aqua equally to the other colors. For all treatments, neither generation nor line show statistically significant effects.
Fig. 2

Each panel depicts the proportion of eggs laid on aqua for each generation of a treatment. The lines depict the replicate populations for each treatment. For panel a and b, the null expectation of no preference is 0.5. For panels c and d, this null expectation is 0.33. For any given population, variance can be high between generations; this is true for any experimental evolution study we have conducted. However what is clear from these graphs is that we do not see the predicted increase over time for the preference for aqua in panels ac. For panel d, we predict equal preference for each color, thus the preference for aqua in this graph should more closely match 0.33, and that does not appear to strongly be the case, although variance does begin to decrease in the final seven generations

Table 2

Comparison of % eggs on aqua between the starting and ending generations (averaged across five generations each)

Treatment

Green

Aqua

Blue

% on aqua generations 1–5

% on aqua generations 21–25

Test 1

Yes

Yes

 

36.6

34.7

Test 2

 

Yes

Yes

44.2

46.6

Test 3

Yes

Yes

Yes

23.6

23.3

Control 1

Yes

Yes

Yes

23.4

23.9

To compare treatments, we used a slightly different measure to account for the differences in the number of choices available to choose from across treatments (two or three). Here we calculated measure comparing observed choice of aqua to an expected choice matching random, specifically (observed egg # – expected egg #)/(expected egg #). We then calculated a mean for each treatment at each generation point and compared these values in a repeated measures ANOVA, with factors of treatment and a repeated measure of generation on each line (Fig. 3). We find no effect of generation (\(\hbox {F}_{24,1056}=8.94,\,\hbox {p}=0.275\)), confirming the individual treatment analyses. We do however find a statistically significant effect of treatment, (\(\hbox {F}_{3,44}=13.73,\,\hbox {p}=0.000002\)), with the aqua/blue treatment significantly differing from every other treatment (all p < 0.00002; Fisher’s LSD). The interaction between generation and treatment is not statistically significant (\(\hbox {F}_{72,1056}=1.07,\,\hbox {p}=0.327\)).
Fig. 3

To directly compare the treatments, we calculated a measure comparing what we observe to our expectation based on how many choices were present. A positive value indicate laying more eggs on aqua than expected, while a negative value indicate the opposite. We find no effect of generation (\(\hbox {F}(24,1056)=1.155,\, \hbox {p}=0.2753\)), confirming our analysis of the replicate populations. We also find no interactions between generation and treatment (\(\hbox {F}(72,1056=0.133,\, \hbox {p}=0.3271\)). Flies in each treatment are laying fewer eggs on aqua than predicted. We do find a significant effect of treatment (\(\hbox {F}(3,44)=13.734, \hbox {p}=0.000002\)). In this case the aqua and Blue treatment differs from all the other treatments (p < 0.00002 for each), which do not differ from each other (Fishers LSD test)

Fig. 4

Results of assays from the end of the experimental evolution study. Performance across the treatments are shown for a blue versus aqua, b aqua versus green. c blue, green, and aqua, d The focal colors plus violet and yellow. Throughout, the control lines, who only experienced plain food during the experimental evolution, match the performance of the baseline population before starting the experiment

3.2 Post-selection assays

We analyzed each post selection assay separately. For the two-choice assays (green/aqua and aqua/blue), we analyzed the proportional choice of aqua (Fig. 4a, b). For the 3 and 5 choice assays (Fig. 4c, d), we conducted a repeated measures ANOVA, with proportional choice of each color as a repeated measure for each line, and treatment as a main effect. For all analyses with proportions, we arcsine transformed the data; the graphs show the original proportions. We find no statistically significant effect of treatment for either the green/aqua or the aqua/blue assays (\(\hbox {F}_{4,55}=1.98,\, \hbox {p}=0.11\) and \(\hbox {F}_{4,55}=2.39,\, \hbox {p}=0.0618\), respectively. Within the aqua/blue assay, the plain food controls are significantly different from each treatment where flies were selected for a color preference (all \(\hbox {p}<0.032\), Fisher’s LSD). For the three choice assay, again treatment is not statistically significant (\(\hbox {F}_{4,55}=1,\, \hbox {p}=0.6850\)), however color is statistically significant (\(\hbox {F}_{2, 110}=12,\, \hbox {p}\le 0.0001\)), with flies significantly preferring green to both aqua and blue (Fisher’s LSD: \(\hbox {p}<0.00001\), and \(\hbox {p}=0.0006\)). The interaction between treatment and color is not significant (\(\hbox {F}_{8,110}=1,\, \hbox {p}=0.3643\))For the five choice assay, we find no statistical significance of either treatment (\(\hbox {F}_{4,55}=2.12,\, \hbox {p}=0.0902\)) or color (\(\hbox {F}_{4,220}=1.94,\, \hbox {p}=0.1042\)), or of their interaction(\(\hbox {F}_{16,220}=1.28,\, \hbox {p}=0.2117\))

3.3 Changes in likelihood of laying eggs on aqua

We document no change in the likelihood that an egg will be laid on aqua. Figure 2 shows the proportional choice of aqua for each replicate population within each test treatment and the control that includes colors. The behavior of the flies is quite variable between populations and between treatments. However, we document no significant increase in the percentage of eggs laid on aqua in any of the tests. If the flies had evolved a taste for aqua, Fig. 2 would show an increased percentage of eggs laid on aqua in the later generations (i.e., 20 and above) as compared to the initial generations.

How does the preference for aqua differ depending on how the choice is framed? To address this we analyzed the (observed − expected/expected) for aqua, and compared it across the four assays. Here we used a repeated measures ANOVA to account for the replicate tests for lines, and nested each line within its treatment. We do not find a statistically significant effect assay type (\(\hbox {F}_{3,360}=1.15,\, \hbox {p}=0.328\)), but we do find an effect of treatment \(\hbox {F}_{4,120}=4.61,\, \hbox {p}=0.0017\)). This effect reflects what is evident in the individual assays graphs: the food controls are different from the experimental treatments where flies made choices during the selections. With contrasts, the food control treatment is different from the other treatments when tested in the context of blue (\(\hbox {F}_{1,120}=10.91, \hbox {p}=0.00126\)), or of blue and green (\(\hbox {F}_{1,120}=8.42, \hbox {p}=0.0044\)).

3.4 Changes in fecundity

While running the selections, we observed an increasing number of eggs being laid by the flies (Fig. 5). Overall the number of eggs laid per population more than doubled over the course of the study. Furthermore, this increase in fecundity occurred all three of the test treatments as well as both control treatments.
Fig. 5

For every treatment we find a significant increase in the rate of laying eggs across the generations of the experiment. Larger numbers of eggs were laid in treatments with more choices. This was analyzed in a repeated measures ANOVA. The effect of treatment is statistically significant (\(\hbox {F}(3,44)=8.94,\, \hbox {p}=0.000097\)). Generation is statistically significant (F(24,1056)=28.81, P<0.000001). The interaction between the two is also statistically significant (\(\hbox {F}(72,1056)=1.7033,\, \hbox {p}=0.000347\)). The equal fitness treatment has a significantly higher number of eggs across the experiment than the other treatments, which do not differ from each other

Because early selections differed slightly in total time allowed for flies, we calculated a measure of the number of eggs laid per minute for each generation of the experiment (Fig. 5). We then analyzed these values in a repeated measures ANOVA, with a main factor of treatment and a repeated measure of generation on each line. Here we find a statistically significant effect of both treatment (\(\hbox {F}_{3,44}=8.94,\, \hbox {p}=0.000097\)), generation (\(\hbox {F}_{24,1056}=28.81,\, \hbox {p}<0.000001\)), and the interaction between the two (\(\hbox {F}_{72,1056}=1.70,\, \hbox {p}=0.000347\)). The rate of egg laying increases with increasing generations, with the highest rates seen in the color control, where choosing each of the three colors was associated equal fitness.

4 Discussion

In this experiment, we selected for eggs laid on the color aqua. The purpose of the experiment was to analyze the evolution of preferences. If the flies evolved in response to selection, we predicted that they would lay more eggs on aqua. There are two, complementary evolutionary paths that flies could take to lay more eggs on aqua. First, the flies could keep their number of eggs laid constant while laying a higher percentage of those eggs on aqua. Second, the flies could increase their overall number of eggs without changing the percentage of eggs laid on aqua. Of course, the flies could use some combination of both of these ways to increase the number of eggs laid on aqua.

When we designed the experiment, we expected the flies to exhibit some change in the first of these two possible changes; we predicted that the flies would evolve to lay a higher percentage of eggs on aqua. This is not what occurred in the experiment. The flies did evolve to lay more eggs on aqua, but not a higher percentage of eggs on aqua; all of the evolution that we observe came in the form of increased fecundity. The flies in this experiment did not evolve a “taste” for aqua in the sense of a higher percentage of eggs on aqua—the flies simply produced more eggs.

Color preference in choosing where to lay eggs did not respond to selection in the way in that we predicted. We found no effects of generation or treatment in the choices the flies made in the predicted directions. Because the lines in each treatment experienced different choices under the selection procedure, we conducted a number of post-selection assays comparing evolved lines to each other and to controls. In the post-selection assays we tested preference across a range of framing contexts (2 choices, 3 choices, and 5 choices). We did find differences when we tested each population in these equivalent assays.

We found a surprising preference for the green hue regardless of the framing of that choice, even in the case where laying eggs on aqua instead of green resulted in fitness. Indeed, the experimental treatments all share similar patterns of preference, which is startling in comparison with the control flies who never experienced an evolutionary history making oviposition choices based on color. These flies demonstrated preferences quite close to our baseline pilot testing of the source population.

All of the flies in the selection treatments did appear to have solved the problem of preferential hue selection not by laying more eggs on aqua, but by increasing fecundity across evolutionary time. In other words they laid more eggs each generation across all the choices presented to them. This egg laying rate increase differed by type of treatment, with flies laying even more eggs in treatments where more substrate choices were present. Fecundity has been shown to evolve in an adaptive manner in prior studies. For example, flies that lay more eggs early die at younger ages (Rose 1984).

Instead of a measurable change in expressed preference, we found that flies responded to the selective pressures of our experimental treatments with increased rates of egg laying instead of a measurable change in a color preference of where to lay eggs. Increased fecundity is predicted in some environmental situations. For instance, in variable environments, the production of more offspring increases the chances of some of the offspring making it to adulthood, and even more so if the number of male partners has increased. More offspring can trade off with having specialized offspring, increasing the mean fitness of offspring in multiple environments or environments of high selective pressure and variance (Fox and Rauter 2003). A recent example (Foucaud et al. 2016) conducted an experiment to specifically calculate fecundity and learning during an invasion by a novel Drosophila subobscura population to a native Drosophila subobscura population. While both species performed roughly the same in cognitive learning abilities, the invasive species had a higher fecundity than the native population.

In our experiment, we observed a steadily increasing rate of egg laying with additional generations, far exceeding any increase in fecundity observed in our past experimental evolution studies on oviposition choice. A likely explanation for our results is that flies are choosing all the options. Flies may have chosen randomly or equally in response to selective pressure. When color preference did not quickly respond to selection, flies laid more eggs, and more eggs even still in treatments with more possible choices. Once females start laying more eggs, the likelihood that an individual egg makes it to the next generation decreases. Females respond by laying more eggs. Essentially we see an arms race of individuals trying to maximize their fitness in a very competitive situation. There are always only 480 eggs collected, but an ever-increasing number of eggs per female. One alternative explanation is that the sex ratio of the populations changed across time, with the number of females was increasing over time and the number of males were decreasing over time. While we did not empirically test this, there was no noticeable sex ratio difference during collecting large single-sex samples for potential future genomics works.

In summary, we do not have a clear explanation for the increase in fecundity that we observe in the experiment. It is possible that some aspect of the experimental setup (in both control and test conditions) was more favorable to the flies. One related possibility is that somehow this environment selected for flies with improved overall health and vigor. Further studies would be needed to determine the cause of the increased fecundity.

When a trait has not appeared to have changed with selective pressure, there is a list of usual possible explanations. It is possible that our starting population did not have adequate genetic variability to allow for color vision preference and evolution of preference. From long-term experimental evolution work in flies, it appears that selection at these time scales acts by shifting distributions of existing variability (Burke et al. 2010). Though our starting population was wild caught flies kept at high population sizes, it always possible that the genetic variability needed for the traits of interest was not present in that population.

It is also possible that too few generations were under selective pressure for behavioral preferences to evolve. In many experiments, the speed of behavioral evolution has been relatively quick. For instance, previous experiments on learning in flies showed significant differences within 8 generations (Mery and Kawecki 2002). In our experiment, some flies in the population, and their progeny, could very well have shown a response to the preference selections, but they never attained the frequencies within the population to see a difference on average. Thus while preference may have evolved to some extent within the population, there was no early selective sweep. Any evolution to lay more eggs on aqua was likely swamped by the increasing fecundity. This could have been because fecundity-related genes were more variability and/or more responsive to changes in selective pressure. The result being that other behaviors may not alter as readily because the one behavior is responding, even if it is not the behavior that we hypothesized we would be selecting on.

Color vision may be very difficult to change. A third possibility that the genes involved in color vision and color preference behavior are pleiotropic, or participate in other behaviors or phenotypic traits. These other traits could be highly conserved and even show redundancy to prevent variations. Additionally, the genes for color vision could be downstream of other activation genes that cannot be altered as readily as fecundity. For instance, the major gene controlling eye morphology is the Pax6 gene, which is found in most animals and insects for controlling eye development and shows little variation (Kozmik 2008). It is well known that many aspects of development can be influenced by single genes such as Pax6 in vision, and, in such a case, selection is not possible without affecting the development and function of other crucial systems (Suzuki et al. 2016; Zhu et al. 2017). Pleiotropic effects among single and series of genes are a commonly recognized constraint on potential evolution of adaptive responses. (Interestingly, some of the early theoretical work in economics (Hansson and Stuart 1992) made an similar argument regarding pleiotropic effects.)

It could be that color vision itself cannot evolve because there is not a selective pressure strong enough to influence it without some additional genetic duplication event first. In some other species, the gain of another level of complexity of color vision- dichromats to trichromats- is due to a duplication even of one of the genes of opsin, and then the specialization of the duplicated gene to another maximum wavelength of light (Dulai et al. 1999; Zhang 2003; Frentiu et al. 2007). That initial redundancy of opsin genes allows for the threshold of signal detection theory to be altered to prevent misses while retaining the original threshold criteria on the original gene.

Finally, the selective behavior in a two- or three- choice paradigm may not be the best conditions for the experimental evolution of color vision in insects. Another task that involves color vision and selective pressure could have produced a more noticeable change in the behavior. It is also possible that fecundity is the only behavioral result that selective pressure on color vision would produce, only because the genes that are involved with the development of color vision are directly involved in the construction and development of other senses.

In conclusion, we performed an experimental evolution study in D. melanogaster where we created selection for laying eggs on aqua. The flies did evolve to lay more eggs on aqua. However, this evolution took place entirely because the flies evolve to lay more eggs overall, but showed no increased ‘taste’ for aqua. This highlights the fact that the evolution of preferences is not necessarily straightforward. This is an initial study, and the results highlight the need to do additional studies on the evolution of preferences both with Drosophila and other species.

Notes

Acknowledgements

Authors would like to thank Ulrich Witt, Editor of the Journal of Bioeconomics who ran the review process for this paper, and two anonymous reviewers, for their excellent comments. We also thank Pamela Tocco, Toni Walker, the students of the Dunlap Lab, and the extended Marcus-Yoakum family. In addition, we would like to thank Itachi Mills for working on an independent verification of the preference and fecundity effects. The work was supported, in part, by NSF Grant: IOS-1021183

References

  1. Abed-Vieillard, D., Cortot, J., Everaerts C., & Ferveur J. F. (2013). Choice alters Drosophila oviposition site preference on menthol. Biology Open.  https://doi.org/10.1242/bio.20136973.
  2. Brembs, B., & de Ibarra, N. H. (2006). Different parameters support generalization and discrimination learning in Drosophila at the flight simulator. Learning & Memory, 13(5), 629–637.CrossRefGoogle Scholar
  3. Brennan, T. J., & Lo, A. W. (2011). The origin of behavior. Quarterly Journal of Finance, 01(01), 55–108.CrossRefGoogle Scholar
  4. Burke, M. K., Dunham, J. P., Shahrestani, P., Thornton, K. R., Rose, M. R., & Long, A. D. (2010). Genome-wide analysis of a long-term evolution experiment with Drosophila. Nature, 467(7315), 587–590.CrossRefGoogle Scholar
  5. Burnham, T. C., Dunlap, A., & Stephens, D. W. (2015). Experimental evolution and economics. SAGE Open, 5(4), 1–17.CrossRefGoogle Scholar
  6. Carfagna, M., & Lancieri, M. (1971). Colour vision and the choice of substrate during oviposition in Drosophila melanogaster Meig. Monitore Zoologico Italiano-Italian Journal of Zoology, 5(4), 215–222.Google Scholar
  7. Collins, J., Baer, B., & Weber, E. J. (2016). Evolutionary biology in economics: A review. Economic Record, 92(297), 291–312.CrossRefGoogle Scholar
  8. Del Solar, E., Guijón, A. M., & Walker, L. (1974). Choice of colored substrates for oviposition in Drosophila melanogaster. Italian Journal of Zoology, 41(3), 253–260.Google Scholar
  9. Del Solar, E., & Ruiz, G. (1979). Behaviour changes in Drosophila melanogaster in the choice of colored substrates for oviposition. Italian Journal of Zoology, 46(1–2), 17–22.Google Scholar
  10. Dulai, K. S., von Dornum, M., Mollon, J. D., & Hunt, D. M. (1999). The evolution of trichromatic color vision by opsin gene duplication in New World and Old World primates. Genome Research, 9(7), 629–638.Google Scholar
  11. Dunlap, A. S., & Stephens, D. W. (2014). Experimental evolution of prepared learning. Proceedings of the National Academy of Sciences, 111(32), 11750–11755.CrossRefGoogle Scholar
  12. Dweck, H. K., Ebrahim, S. A., Kromann, S., Bown, D., Hillbur, Y., Sachse, S., Hansson, B. S., Stensmyr, M. C. (2013). Olfactory preference for egg laying on citrus substrates in Drosophila. Current Biology, 23(24), 2472–2480.Google Scholar
  13. Erclik, T., Hartenstein, V., McInnes, R. R., & Lipshitz, H. D. (2009). Eye evolution at high resolution: The neuron as a unit of homology. Developmental Biology, 332(1), 70–79.CrossRefGoogle Scholar
  14. Foucaud, J., Moreno, C., Pascual, M., Rezende, E. L., Castaneda, L. E., Gibert, P., Mery, F. (2016). Introduced Drosophila subobscura populations perform better than native populations during an oviposition choice task due to increased fecundity but similar learning ability. Ecology and Evolution, 6(6), 1725–1736.Google Scholar
  15. Fox, C. W., & Rauter, C. M. (2003). Bet-hedging and the evolution of multiple mating. Evolutionary Ecology Research, 5(2), 273–286.Google Scholar
  16. Frentiu, F. D., Bernard, G. D., Sison-Mangus, M. P., Van Zandt Brower, A., & Briscoe, A. D. (2007). Gene duplication is an evolutionary mechanism for expanding spectral diversity in the long-wavelength photopigments of butterflies. Molecular Biology and Evolution, 24(9), 2016–2028.CrossRefGoogle Scholar
  17. Gonzalez-Bellido, P. T., Wardill, T. J., & Juusola, M. (2011). Compound eyes and retinal information processing in miniature dipteran species match their specific ecological demands. Proceedings of the National Academy of Sciences, 108(10), 4224–4229.CrossRefGoogle Scholar
  18. Hansson, I., & Stuart, C. (1990). Malthusian selection of preferences. American Economic Review, 80(3), 529–544.Google Scholar
  19. Hansson, I., & Stuart, C. (1992). Socialization and altruism. Journal of Evolutionary Economics, 2(4), 301–312.CrossRefGoogle Scholar
  20. Harzsch, S., Melzer, R. R., & Müller, C. H. (2007). Mechanisms of eye development and evolution of the arthropod visual system: The lateral eyes of Myriapoda are not modified insect ommatidia. Organisms Diversity & Evolution, 7(1), 20–32.CrossRefGoogle Scholar
  21. Heisenberg, M., & Wolf, R. (2013). Vision in Drosophila: Genetics of microbehavior. New York: Springer.Google Scholar
  22. Joseph, R. M., Devineni, A. V., King, I. F., & Heberlein, U. (2009). Oviposition preference for and positional avoidance of acetic acid provide a model for competing behavioral drives in Drosophila. Proceedings of the National Academy of Sciences, 106(27), 11352–11357.CrossRefGoogle Scholar
  23. Kahneman, D., Knetsch, J., & Thaler, R. H. (1991). Anomalies: The endowment effect, loss aversion, and status quo bias. The Journal of Economic Perspectives, 5(1), 193–206.CrossRefGoogle Scholar
  24. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–292.CrossRefGoogle Scholar
  25. Kozmik, Z. (2008). The role of Pax genes in eye evolution. Brain Research Bulletin, 75(2), 335–339.CrossRefGoogle Scholar
  26. Mery, F., & Kawecki, T. J. (2002). Experimental evolution of learning ability in fruit flies. Proceedings of the National Academy of Sciences, 99(22), 14274–14279.CrossRefGoogle Scholar
  27. Miller, P. M., Saltz, J. B., Cochrane, V. A., Marcinkowski, C. M., Mobin, R., & Turner, T. L. (2011). Natural variation in decision-making behavior in Drosophila melanogaster. PLoS ONE, 6(1), e16436.CrossRefGoogle Scholar
  28. Panchanathan, K., & Boyd, R. (2003). A tale of two defectors: The importance of standing for evolution of indirect reciprocity. Journal of Theoretical Biology, 224(1), 115–126.CrossRefGoogle Scholar
  29. Peterson, K. J., & Eernisse, D. J. (2016). The phylogeny, evolutionary developmental biology, and paleobiology of the Deuterostomia: 25 years of new techniques, new discoveries, and new ideas. Organisms Diversity & Evolution, 16(2), 401–418.CrossRefGoogle Scholar
  30. Robson, A. (1994). The evolution of attitudes to risk: Lottery tickets and relative wealth. Mimeo: University of Western Ontario.Google Scholar
  31. Robson, A. J., Szentes, B., & Iantchev, E. (2012). The evolutionary basis of time preference: Intergenerational transfers and sex. American Economic Journal: Microeconomics, 4(4), 172–201.Google Scholar
  32. Rogers, A. R. (1994). Evolution of time preference by natural selection. American Economic Review, 84(3), 460–481.Google Scholar
  33. Rose, M. (1984). Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution, 38, 1004–1010.CrossRefGoogle Scholar
  34. Rubin, P. H., & Paul, C. W. (1979). An evolutionary model of tastes for risk. Economic Inquiry, 17(4), 585–596.CrossRefGoogle Scholar
  35. Salcedo, E., Huber, A., Henrich, S., Chadwell, L. V., Chou, W.-H., Paulsen, R., Britt, S. G. (1999). Blue-and green-absorbing visual pigments of Drosophila: Ectopic expression and physiological characterization of the R8 photoreceptor cell-specific Rh5 and Rh6 rhodopsins. Journal of Neuroscience, 19(24), 10716–10726.Google Scholar
  36. Salomon, C. H., & Spatz, H.-C. (1983). Colour vision in Drosophila melanogaster: Wavelength discrimination. Journal of Comparative Physiology A: Neuroethology, Sensory, Neural, and Behavioral Physiology, 150(1), 31–37.CrossRefGoogle Scholar
  37. Schnaitmann, C., Garbers, C., Wachtler, T., & Tanimoto, H. (2013). Color discrimination with broadband photoreceptors. Current Biology, 23(23), 2375–2382.CrossRefGoogle Scholar
  38. Stigler, G. J., & Becker, G. S. (1977). De Gustibus Non Est Disputam. The American Economic Review, 67(2), 76–90.Google Scholar
  39. Suzuki, T., Takayama, R., & Sato, M. (2016). Eyeless/Pax6 controls the production of glial cells in the visual center of Drosophila melanogaster. Developmental Biology, 409(2), 343–353.CrossRefGoogle Scholar
  40. Tversky, A., & Kahneman, D. (1974). Judgement under uncertainty: Heuristics and biases. Science, 185(4157), 1124–1131.CrossRefGoogle Scholar
  41. Washington, C. G. (2010). Color vision in Drosophila melanogaster. New York: Columbia University.Google Scholar
  42. Yang, C.-H., Belawat, P., Hafen, E., Jan, L. Y., & Jan, Y.-N. (2008). Drosophila egg-laying site selection as a system to study simple decision-making processes. Science, 319(5870), 1679–1683.CrossRefGoogle Scholar
  43. Zhang, J. (2003). Evolution by gene duplication: An update. Trends in Ecology & Evolution, 18(6), 292–298.CrossRefGoogle Scholar
  44. Zhu, J., Palliyil, S., Ran, C., & Kumar, J. P. (2017). Drosophila Pax6 promotes development of the entire eye-antennal disc, thereby ensuring proper adult head formation. Proceedings of the National Academy of Sciences, 114(23), 5846–5853.CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2017

Authors and Affiliations

  • Mellissa Marcus
    • 1
  • Terence C. Burnham
    • 2
  • David W. Stephens
    • 3
  • Aimee S. Dunlap
    • 1
  1. 1.Department of BiologyUniversity of Missouri, St. LouisSt. LouisUSA
  2. 2.Argyros School of Business and EconomicsChapman UniversityOrangeUSA
  3. 3.College of Biological Sciences, University of Minnesota, Twin CitiesMinneapolisUSA

Personalised recommendations