Mike Ko Personal Portfolio


Home-School Education
Hong Kong

University of Durham
Bachelor of Science
United Kingdom

University of Sussex

Master of Arts
United Kingdom

Academic Works
                                Year 3
: Research Project

Assessing the Potential Relationship of Male Exploratory Behaviours and Body Colour on Mate Choice by Females in Guppies (Poecilia reticulata)


            Mate choice by females can be influenced by a variety of factors, including body colour and male behaviours. By measuring proxies for female mate choice as well as boldness and exploratory behaviour in males, the potential correlation between these variables as well as colour was examined in Poecilia reticulata. Mate choice was assessed by the duration of female positional affiliation with two associated males. Colour match between the males and females were also noted for each mate choice tests. Male boldness was evaluated through male proximity to a novel object both in isolation and in groups and was recorded as proximity scores. Male exploratory behaviour was gauged by extent of travel, measured in grids, in the presence of a novel object while in isolation. Colour match was significantly correlated with mate choice, which agrees with findings from previous studies. Proximity scores for males in groups were correlated with mate choice, but scores for males in isolation was not. This potentially indicates sexual selection on boldness only under social conditions. Although useful, results obtained from proxy variables should be interpreted with care. Future studies can potentially investigate how multiple factors influence mate choice simultaneously.


           Sexual selection is a crucial selective pressure in evolution, in particular for animals. For a given species, individuals of either sex have different levels of mating successes and general fitness. Thus both sexes will often be under selective pressures to choose comparatively fit mating partners. In many cases, females make a greater investment towards reproduction and thus are more selective in their mate choice of male mates (Darwin, 1871). Such non-random mate choice can account for the development of various morphologies, behaviours or other phenotypes. As such, understanding how given factors affect mate choice can provide insights into how given traits contribute to fitness. More practically, such knowledge can allow better predictions of mating success, which can be useful in improving conservation efforts or in projecting population levels.

           Mate choice can be influenced by a variety of factors. The effect of colour on mate choice by females has been well studied in fishes. Experiments on cichlids suggested that colour is a determinant factor that leads to assortative mating, such that females tend to mate with males of similar colour (Seehausen and van Alphen, 1998; Maan and Sefc, 2013; Selz et al., 2016). Others have found similar evidences in darters (Williams and Mendelson, 2011) and medaka (Utagawa et al., 2016). In guppies, melanistic (spotted) females have also been found to prefer similarly melanistic males (Culumber et al., 2014). Such preferences may not be permanent, as female guppies have been shown to prefer males that have dissimilar colouration relative to the previous mate (Eakley and Houde, 2004).

           Non-morphological factors like behaviours may also influence mate choice. In particular, female mate choice can be affected after exposure to males exhibiting certain behaviours. Relatively exploratory zebra finch females show a preference for males that have been observed to be most exploratory (Schuett, Godin and Dall, 2011). Similarly, female guppies were found to choose bolder males regardless of colour (Godin and Dugatkin, 1996). In Siamese fighting fish, females prefer males that exhibit less aggressive behaviours (Dzaieweczynski et al., 2014). Additionally, the surrounding social environment can further affect mate choice. Under experimental conditions where male-male interactions were prevented, female damselfish demonstrated mate preference based on male colour. Yet when male-male interactions were allowed, the females showed no mate preference (Wacker et al., 2016).

           The guppy Poecilia reticulata is a model organism native to certain South American and Caribbean countries and inhabits warm fresh or brackish waters (Global Invasive Species Database, 2017). They are often used in genetic and mating studies due to its sexual dichromatism and distinct colour polymorphisms in males (Endler, 1978; Godin and Dugatkin, 1996).

           This study tested whether indirect proxies for male exploration or boldness behaviours have any relationship with mate choice exhibited by females in P. reticulata. Mate choice by females was first assessed, followed by separate evaluations of male exploratory or boldness behaviour, the latter both in both isolation and in conspecifc groups. In addition, the effect of colour on mate choice was also analysed.


           Experiments were carried in a laboratory at the Department of Biosciences (54°45'54.0"N 1°34'20.4"W), University of Durham from approximately 9:00 to 17:00, between 16 February to 3 March 2017. Sample population of P. reticulata guppies were obtained at random and consisted of 10 females and 20 males. Males were further separated into Set A and Set B (ten fishes each) and assigned a set-number code (A1, B10 etc.). Body colours of the guppies included blue, yellow, red and bronze, with each being the dominant non-silver colour on each guppy.

           Guppies were kept in three “maintenance” glass tanks with de-chlorinated water and heaters to maintain a water temperature of approximately 26 °C. Females were kept in a medium tank (46x19x30 cm) while the two male groups were each kept in large tanks (61x31x30 cm). The sides and back of the tanks were visually shielded with blue paper to minimise disturbance. Tanks were filter-cleaned in the morning, five times per week. Feeding of approximately ten food pellets per group was done twice a day, once in the morning and late evening. Guppies were only fed once over weekends with no tank cleaning.

Experiment 1: Mate Choice, Exploration and Boldness

           Randomly chosen females were grouped with one male from Set A and Set B each to form 31 three-fish test groups. Each test group were subjected to a three tests: 1. first mate choice test; 2. isolated exploratory-boldness test and; 3. second mate choice test.

           The first mate choice test was done on each test group using a setup with two adjacent small glass tanks (31x20x20 cm) (Figure 1A). The two males of each test group were placed into separate compartments formed by an black opaque partition in the “Far’ tank. The associated female was placed “Near” tank, with a clear view of the separated males in the adjacent tank. The guppies were left alone for ten minutes to acclimatise. Next, recordings were made over ten minutes on the total elapsed time the female spent on the left and right sides of her tank, which were here taken as time spent affiliating with either male. The time difference was calculated and used as an indication of mate choice exhibited by females. Specifically, the male on the side where the female stayed the longest was regarded as preferred. Whether the female colour matched with that of either of the two males was also recorded, specifically through the following case notation: ‘0’ for no match with both males, ‘1’ for match with the less preferred (‘losing’) male, and ‘2’ for match with the more preferred (‘winning’) male.

           The pair of males from each test group was then transferred to other glass tanks for the following isolated exploratory-boldness test. Each male was placed into two separate but adjacent glass tanks of similar size to those in the mate choice test. Shielding with blue paper was done on all sides except for the one facing the observer. For each tank, a grid of 15 by 9 squares (2x2 cm per square) was laid underneath and a novel object was placed inside at one end (Figure 1B). The novel objects were stacks of three doughnut-shaped discs with an alternating grey-white-grey colour pattern. Before each test, the side of the tanks with the novel object was hidden by a black opaque partition, and males are placed on the other side. After an acclimatisation period of five minutes, the partitions were removed and the positions of the two males were simultaneously tracked over ten minutes using a digital video recorder positioned above both tanks. An Excel Visual Basic for Application program was used post-hoc to note the grid position of each male every 30 seconds for the first five minutes of each video recording. The program assigned each grid an associated score based on its proximity to the novel object. These “Isolated Proximity” scores were summed for each male and were taken as proxies for male boldness. Similarly, the number of unique grids occupied over the same observation period was used to create a “Unique Grid” score to indicate exploratory tendency. The two males were transferred back to their tank compartments in the mate choice test. After a ten-minute acclimatisation period, the second mate choice test was carried out in the same manner as the first test.

Figure 1. Experiment 1 setup for A) Mate Choice test and B) Exploratory-Boldness test. Red lines denote sides where blue paper covered with blue paper as visual shielding. All tanks have a heater to cylindrical heater to maintain temperature. Blue arrow in A) denotes black partitions that were not present during tests. Purple arrows in B) show the partitions that are removed at the start of each 10-minute observation period.

Experiment 2: Boldness in Social Environment

           The two sets of male guppies were kept in their respective maintenance tanks when testing for male boldness in the presence of conspecifics. White backgrounds were added behind each tank, with a seven-zone grid was placed near one end (Figure 2). A digital video recorder was set up on a tripod about one meter away from the unshielded side of the tanks to record guppy positions. Before each test, blue paper shielding on the gridded end of the tank were temporarily removed before each test, followed by a five-minute acclimatisation period. Novel objects were then placed about five centimetres away from the gridded end of the tank. Video footages for recording guppy proximity to the novel objects were taken over ten-minute periods, after which shielding was put back into position. This process was done in an alternating fashion between the two sets of males for each novel object. Objects included plastic fish models, a black plastic bottle, and moulded clay of various shapes and colours. A total of 18 videos recording proximity to 17 different novel objects (some used more than once) were made for each of the two male sets. For each video footage, the proximity of each male using the seven-zone grid were recorded post-hoc every 30 seconds over the first five minutes. The zone numbers of the resulting data were then used as a linear proximity score and summed to give the subtotal score for the five-minute period. Each male’s score subtotals from all video observations were then summed to give a total Group Proximity score for the given male. These scores were used as proxies for boldness in the presence of other males.

Figure 2. Setup in maintenance tank with Set A of male guppies for testing male boldness whilst in a group. Each of the seven zones consisted of two columns on the grid behind the tank. Zone 0 corresponds to everywhere else beyond the grid. Novel objects were placed just outside the left side of the tank. The whole setup is mirrored in the tank where Set B of males was tested.

Statistical Tests

           Three statistical tests were done for this study using the IBM SPSS Statistics 22.0 software. Pearson's chi-squared test was used on the colour match data to evaluate whether there were significant relationships between colour and test groups. Because all test groups had less than five counts for colour match, the output for the Fisher’s Exact Test was used to evaluate significance. Spearman’s ranked correlation coefficient test was applied to the Isolated Proximity and Unique Grid scores from Experiment 1 to test for potential correlation. The univariate general linear model was used to determine if colour match, Unique Grid and Group Proximity scores from Experiment 2 had an effect on mate choice. The Shapiro-Wilk test was used to assess the normality of the data on the time difference variable from mate choice tests.


           Females in 62 female mate choice tests done showed varying levels of mate preferences. There were no clear trends in time differences in female affiliation with males and thus mate choice (Figure 3A). Time differences can range from less than a minute (no preference) to ten minutes. More than half (37) of mate choice tests had time differences of less than five minutes. Only 7 out of 29 (24%) test groups had females that preferred different males between the first and the second mate choice test. The Shapiro-Wilk test on time difference in female affiliation showed that the data was normally distributed (df = 61, p = 0.101)

           Almost half of the test groups had females with colours that did not match either of the associated two males. For the remaining tests, more females preferred males with matching colour (Figure 3B). Analysis with the Pearson's chi-squared test supported this result (p = <0.01), showing that there is a significant relationship between colour match and time difference in female affiliation. One mate group’s female spent equal times in both sides of the tank, thus colour match could not be assessed.

Figure 3. A) Histogram for the time difference between the time females stayed in the two sides of the “Near Tank” during mate choice tests. Negative time difference values indicates preference for Set A males, while positive values indicate preference for Set B males. B) The proportion of mate choice test results with the three cases of colour-match between female and male guppies.

           In Experiment 1’s isolated exploratory-boldness test, the Unique Grid score of males ranged from 7 to 42 grids, while Isolated Proximity scores were between 193 and 659. Medians were 25 grids and 474, respectively (Figure 4).

           At first glance, Isolated Proximity and Unique Grid scores did not show a significant correlation. Yet if a group of potential outlier data points (indicated by black arrow in Figure 5) were ignored, the majority of the data suggested an inversely proportional relationship between the two variables. Further analysis with the Spearman’s Rank Correlation Coefficient showed that there was indeed a negative correlation (rs = -0.538, p = <0.001).

Figure 4. Box plots of data values obtained from the isolated exploratory-boldness tests on male guppies in Experiment 1: A) Number of Unique Grids Entered by Males; B) Isolated Proximity Score.

Figure 5. Scatter plot of data points on Isolated Proximity and Unique Grid scores. Black arrow denotes data cluster that are potential outliers.

           Analysis with the univariate general linear model showed that time differences in female affiliation to males had no significant relationship with Unique Grid scores (F = 0.615, p = 0.437). However, there was some evidence for significant relationships with both Group Proximity scores (F = 4.808, p = 0.033) and Colour Match (F = 6.547, p = 0.47).


           This study primarily investigated whether mate choice made by female guppies was influenced by male exploratory behaviour and boldness in isolated and social environments. Most notably, the case distribution of colour match between males and females was significantly correlated to mate choice. The proxies for male exploration and boldness, Unique Grids and Isolated Proximity scores, were also correlated. Also, statistical test results suggested that female mate choice was significantly correlated to male boldness exhibited when males were in social conditions but not when in isolation.

           Relatively few females changed their male affiliation between the first and second mate choice test in Experiment 1. The result suggests that at least certain female guppies exhibited non-random mate preferences, which may be based on colour. Significant results from two statistical tests further supports this notion and agrees with existing work in the literature (Seehausen and van Alphen, 1998; Maan and Sefc, 2013; Selz et al., 2016), although the relationship found in this study was relatively weak. This partial discrepancy may possibly be accounted for by the small sample size and a relatively wide diversity in guppy body colours. The latter resulted in generally incomplete colour matches between males and females. More conclusive relationships might be found if a sample population with a less random diversity of body colours were used.

           The negative correlation found between males’ Unique Grid and Isolated Proximity scores was a reasonable result. The more a given male explores his surroundings, the greater the chance that it will move away from a given point (i.e. novel objects). Given their correlation, only Unique Grid scores were used in the final Univariate test to simplify analysis.

           Previous studies have shown that boldness or exploratory behaviour can be both positively (Godin and Dugatkin, 1996; Schuett, Godin and Dall, 2011) or negatively correlated (Dzaieweczynski et al., 2014) to female mate choice. Thus it is interesting that the Unique Grid and Isolated Proximity scores were not significantly correlated to mate choice. Yet together with the fact that mate choice was correlated with Group Proximity scores, the result might perhaps reflect on-going sexual selection. For example, females may tend to avoid male mates who demonstrate bold behaviour when alone, which can be potentially be disadvantageous due to increased risk of predation.

           It should be noted that the effect of male behaviour on mate choice could only be indirectly inferred for this study. In particular, the female guppies have not directly witnessed the relevant male behaviours. This was not done in this study due to limitations in available equipment. It was thus an implicit but not necessarily true assumption that males’ exploration-boldness behaviours can manifest itself through other means to females during the mate choice tests. As such, there may actually be other male-based or even external factors that influenced mate choice. Further studies can potentially infer more direct relationships between variables by allowing females observe males while the latter’s exploratory behaviour or boldness is measured.

           Perhaps a more interesting question is to compare the level of influence between different factors on mate choice. It is probable that mate choices made by female guppies or other species are influenced by multiple factors simultaneously, some more than others. For example, Wacker et al., (2016) found that both body colour and observed male-male interactions affected mate choice in coral reef fish. Yet the latter was the more decisive factor when both were acting simultaneously. Thus there may be an inherent hierarchy of factors that ultimately determine mate choice. Some factors that may be regarded as competing influences can potentially be linked. Schweitzer, Motreuil and Dechaume-Moncharmont (2015) found that bolder convict cichlids often had darker colouration, which can then affect female mate choice as a whole.

            Caution should be exercised in interpreting this study’s results due to a number of assumptions. In the mate choice test, female occupation of a side of a tank in front of a given male was assumed to signify attraction. Yet this may not be the case, as other factors such as organic debris (e.g. faeces) can potentially have lured the female to a specific side. One study on cichlids has shown that female affiliation near compartments with males did not correspond to mate preference during actual courtship or spawning interactions (Egger et al., 2008).  Similarly, the Unique Grid, Isolated Proximity and Group Proximity scores might not fully reflect male exploratory behaviour and boldness. Thus sole use of such proxies may potentially overestimate the effect of the underlying factor.


            The author would like to thank Dr. Twiss and everyone at the lab who has been involved in facilitating this project and for tolerating potential student incompetence on the author’s part.



Back to Top