Brain size does not affect reproductive behaviour in male guppies (Poecilia reticulata)
Simon Eckerström Liedholm
Table of Contents
Abstract ... 1
Introduction ... 2
Materials and methods ... 5
Study system ... 5
Experimental setup ... 5
Scoring of behaviour ... 7
Statistical analysis ... 7
Results ... 8
Discussion ... 12
Conclusions ... 16
Acknowledgement ... 17
References ... 17
Abstract
The processes and mechanisms that govern brain size evolution remain a widely discussed topic in evolutionary biology. How relative brain size relates to animal behaviour and cognition is even more controversial. Recent comparative and experimental studies have shown a positive relationship between relative brain size and complexity of behaviour. Some of the most important behaviours that have direct consequences for an individual’s fitness are reproductive behaviours, and they sometimes require quite complex behavioural repertoires.
Selection for complex behaviour might therefore induce an expansion of brain size to allow for cognitively demanding tasks during courtship and mating. In the present study we investigated the effect of relative brain size on reproductive behaviour in male guppies (Poecilia reticulata), using fish from a recently established brain size artificial selection experiment. Females were paired with either a large- or a small-brained male, and we collected data on a suite of male courtship behaviours including sneak copulation attempts, courtship display, gonopodial swings and time spent following the female. Although the extent of orange colouration, a trait that varies across large- and small-brained males, affected male behaviour, we were not able to detect any difference in reproductive behaviour between the brain size selection lines. These results suggest that there is no strong association between male mating behaviour and relative brain size, and future studies will examine this question further. But currently, our results indicate that relative brain size might not be linked to reproductive behaviour to any significant extent, at least not in the guppy.
Introduction
Brain size is a highly variable trait among vertebrate taxa (Northcutt 2002, Striedter 2005).
Some of the variation is explained by the fact that larger animals need relatively larger brains to control their larger bodies, although there is not a perfect 1:1 scaling of body size to brain size ratio (Jerison 1973). Interestingly, also after controlling for allometric scaling rules, there is still substantial variation in brain size left that requires explanation in order to understand the factors that drive brain evolution (Jerison 1973, Roth and Dicke 2005).
Excess brain tissue probably incurs a non-trivial metabolic cost (Mink et al. 1981, Isler 2011), and will consequently be opposed by selection. Therefore, a comparison of relative brain size between or within species might uncover differences in brain function. Since the size of certain brain substructures have been found to positively correlate with their functional performance (Huber and Rylander 1992, Sherry et al. 1992, Devoogd et al. 1993, Maguire et al. 2000), the same relationship can be hypothesized to exist between relative brain size and complexity of behaviour (Sol et al. 2005, Sol et al. 2008). This is because complex behaviours arguably demand considerable cognitive processing power (Striedter 2005, Shettleworth 2010). A positive relationship between complexity of behaviour and relative brain size has been found in several phylogenetic comparative studies. These include song learning and migration efficiency in birds (Garamszegi et al. 2005, Møller 2010), tool use in birds and primates (Lefebvre et al. 2002, Reader and Laland 2002), foraging flexibility in bats (Ratcliffe et al. 2006) and innovativeness in primates (Reader and Laland 2002). Relative brain size has been linked to speed of decision-making in guppies, where individuals with smaller brains seem to make hastier decisions (Burns and Rodd 2008). A recent experimental study, using artificial selection for brain size, found a positive effect of relative brain size on numerical learning in guppies (Poecilia reticulata) (Kotrschal et al. 2013). Furthermore, both group size and deception rate in primates seem to be positively correlated with neocortex size (Sawaguchi and Kudo 1990, Byrne and Corp 2004), which indicate that socially complex behaviour might require enhanced cognitive capacity (Dunbar 1998).
Reproductive behaviours of animals are very variable, and have intrigued scientists ever since Darwin (Darwin 1871). Many of the reproductive behaviours performed by animals are complex (Andersson 1994, Crockford et al. 2007, DuVal 2007, Brown et al. 2012, Bierbach et al. 2013) and might demand considerable cognitive abilities. It has also been suggested that cognition might be under sexual selection through mate choice (Keagy et al. 2009, Boogert et al. 2011b), and that mate choice itself can be cognitively demanding (Keagy et al. 2009, Ryan et al. 2009). The link between reproductive behaviour and cognition has not been extensively investigated, but a few studies suggest a possible relationship. Song learning in birds, for instance, has been suggested to be cognitively demanding (Shettleworth 2010), mainly because of its direct dependence on learning and general brain function (Catchpole 1996, Peters et al. 2014). A positive relationship has been found between song complexity and
study on bowerbirds, males of different species were subjected to cognitively demanding tasks. A positive relationship was then detected when correlating the score for each species with bower complexity (Keagy et al. 2012). Other studies, which have used relative brain size as a proxy for cognitive ability, have found that variation in relative brain size and brain structure is sometimes linked to differences in reproductive behaviours. For instance, a larger brain has been associated with nest building, courting and parental care in males of the three- spined stickleback (Gasterosteus aculeatus) (Kotrschal et al. 2012), mating strategy in brown trout (Kolm et al. 2009), bower complexity in bowerbirds (Madden 2001), song learning in birds (Jacobs 1996), spatial learning in rodents (Jacobs 1996) and mating success in primates (Pawlowski et al. 1998). The degree of polygamy has been found to be negatively correlated with relative brain size in bats (Pitnick et al. 2006), Tanganyikan cichlids (Pollen et al. 2007) and primates (Schillaci 2006). In the study by Pitnick et al. (2006) they also found that testis size increased with degree of polygamy. This observation fits well with the expensive tissue hypothesis (Aiello and Wheeler 1995); since the brain is considerably energetically demanding, an increase in size of the similarly energetically demanding testes leads to a reduction in brain size. Studies on brain gene expression in relation to reproductive behaviour have also been conducted. Aubin-Horth et al. (2005) could demonstrate clear differences in brain gene expression between males of the sneaking mating tactic and males of the courting mating tactic of Atlantic salmon (Salmo salar). Exactly how relative brain size and brain structure differences translate into cognitive ability is not completely resolved however (Byrne and Corp 2004, Iwaniuk and Hurd 2005, Chittka and Niven 2009), and has to be interpreted with possible ecological factors in mind. In general, few studies have investigated this topic, and the relationship between reproductive behaviour, cognition and relative brain size is still unclear.
The studies that have been described so far are mostly comparative. Such studies are highly informative as starting points for novel hypotheses and for investigations of macroevolutionary patterns, but they have a potential weakness in that they cannot be used to determine causality since they are correlational in their nature (Healy and Rowe 2007). In order to determine causality concerning the link between relative brain size and reproductive behaviour, experimental manipulation of brain size is needed. This has been done previously, but differences in reproductive behaviour were not studied (Wimer and Prater 1966, Fuller 1979). Several guppy brain size selection lines have been established recently (Kotrschal et al. 2013), and studying them allows for a more direct approach to understanding the relationship between relative brain size and reproductive behaviour. Fish brains are similar to the brains of other vertebrates (Braithwaite 2005), and guppies are easy to rear in the lab (Houde 1997). We now use these selection lines, with an 11 % difference in relative brain size between the up- and the down-selected lines, to investigate the association between relative brain size and reproductive behaviour.
The guppy is a small (1.5-6 cm) live bearing fish, native to the river systems of Trinidad.
Males are considerably smaller than females, and more colourful. The male colouration is highly variable and usually includes orange, black, yellow and iridescent features. Courtship is complex in the guppy and male guppies have two ways of achieving a copulation; either
through sneak copulation behaviour or through courtship display whereby mating is carried out with female consent (Houde 1997). Both types of reproductive behaviours are often performed at the same time, but the relative preference for one or the other is dependent partly on the environmental conditions (Magurran and Seghers 1990, Magurran and Nowak 1991, Reynolds et al. 1993, Godin 1995, Chapman et al. 2009). Because of this, they are sometimes considered to be alternative reproductive tactics within a single conditional strategy; it is defined as a conditional strategy because the choice of tactic is made in every situation according to the prevailing environmental conditions (Gross 1996).
In the current study, we wanted to determine the effects of relative brain size on the choice of mating tactic in the guppy. The sneaky mating tactic is best performed without the female’s knowledge. It has been suggested that sneak copulation behaviour might be cognitively demanding since it requires careful timing with female behaviour to be effective (Houde 1997). A positive relationship between relative brain size and sneaky behaviour has been found in brown trout (Salmo trutta), although other ecological differences may contribute to the observed differences (Kolm et al. 2009). It has also been suggested that the sneaky mating tactic performed by male sailfin mollies (Poecilia latipinna) requires higher cognitive abilities, owing to the up-regulation of genes involved in learning and memory in the males that perform the sneaky tactic (Fraser et al. 2014). The mating tactics of guppies are very similar to those of sailfin mollies, and it is tempting to think that the findings of Fraser et al.
(2014) might apply to guppies as well. At the same time, ‘normal’ male guppy courtship display might also be cognitively demanding, as it also requires both timing and coordination to be effective (Houde 1997).
Trying to predict the effect of relative brain size on reproductive behaviour using the guppy brain size selection lines is complicated by the fact that other traits than relative brain size differ between the two selection lines. Gonopodium length, tail length and area of orange colouration are all significantly greater in the large-brained lines (Kotrschal et al. in review), and these traits have previously been linked to differences in male reproductive behaviour in wild populations. Males with longer tails have been found to perform sneak copulation attempts more often and courtship display less often (Karino and Kobayashi 2005, Karino and Kamada 2009). Gonopodium length has been found to correlate positively with sneak copulation attempts (Reynolds et al. 1993). In addition, females seem to prefer longer gonopodia, even when controlling for colouration of the male (Brooks and Caithness 1995).
Males with larger areas of orange colouration seem to employ the sneak copulation tactic more often than duller males (Kiritome et al. 2012). Overall, the prediction is that large- brained males will perform more of the behaviours associated with sneak copulation and less courtship displays. This is because they have longer gonopodia, longer tails, they are more colourful, and probably have greater cognitive ability (Kotrschal et al. in review).
Materials and methods Study system
We investigated the effect of relative brain size selection on reproductive behaviour in laboratory reared male guppies that have been artificially selected for large and small relative brain size. The guppies used descended from wild guppies collected in Trinidad in 1998. The large and small brain size populations were established in 2010 through divergent selection on relative brain size according to the following protocol. Females and males were isolated in random pairs and split into three independent replicate groups. Every replicate group consisted of 75 breeding pairs (the F0 generation). The offspring of the breeding pairs were transferred to separate holding tanks and were subsequently split into males and females as soon as sex determination was possible. After the F0 pairs had reproduced they were sacrificed, to be able to measure their brain size. The brain weight and body weight was measured for each individual and their brain weight was regressed onto their body weight.
The offspring of the 15 pairs that had the largest summed relative brain size (controlling for body size), and the offspring of the 15 pairs that had the smallest summed relative brain size were isolated for each replicate and used as breeders for the next generation. These six groups now constituted the F1 generation large- and small-brained lines, respectively. From each of these F1 generation groups, two males and two females of each family were isolated for breeding and paired randomly, avoiding full siblings. The process was then repeated to generate the F2, F3 and F4 generations. After two generations the mean difference in relative brain size between the selection lines was 9 % (p<0.0001, Kotrschal et al. 2013). The fish used in this experiment were fish from the fourth generation with a mean difference in relative brain size of 11 % between large and small brain size selection lines.
Both females and males were kept under a 12:12 l/d lighting condition. The room temperature was held at 26-27°C and the fish were fed flake food 6 days a week before and during the whole experiment. The fish were kept separately in single holding tanks during the experimental period.
Experimental setup
We used an experimental tank that was 19x28x10 centimetres in size for the behavioural observations of reproductive behaviour (fig. 1). Video recordings were made from both the horizontal view and the top view. We used a Sony HDR-SR11 video camera mounted on a tripod for the horizontal view, and a Logitech HD Webcam C615 attached to a retort stand for the top view. The males in the current experiment were taken from the fourth generation of the selection experiment; females were taken from the unselected, freely mating descendants of the founder population. We used already mated females because male guppies adjust their mating behaviour in accordance with the female’s response, and perform more sneak copulation attempts towards previously mated, less receptive females (Farr 1980, Guevara- Fiore et al. 2009). Thus, to increase our chances of observing all possible mating behaviours, including sneak copulation attempts, we used previously mated females in all trials.
We used 10 individuals from each replicate line. Since there were three replicate lines for each brain size selection regime, 30 males were used from each of the brain size selection regime. In total 60 males were used, and they were paired with 60 females in the trials. The experiments were performed during two weeks, with each replicate-brain size combination being represented by one male each day (the behaviour of a total of 6 males was thus collected each day). To avoid biases by the time of day on reproductive behaviour, the order of the males within each day was randomized across the replicate-brain size combinations.
Prior to video recording of behaviour, the pair was gently placed in the tank at the same time.
Every pair was recorded with video cameras from two angles for 25 minutes, and all experiments were conducted between 08:00 am and 12:00 pm (noon) each day.
Figure 1. Schematic illustration of the experimental setup.
Standard length (length of fish excluding tail length) was collected for all males and females immediately after the behavioural experiments. Area of the orange spots were also measured for all males after the experiments, and analysed quantitatively using ImageJ v. 1.44 (Schneider et al. 2012). The size of the orange spots was divided by the standard length of the fish, to get a measure of orange colouration that was independent of the size of the fish. To measure the area of orange colouration, the males were anesthetized in benzocaine for a few seconds before photo capture. They were then placed next to a ruler and a X-rite ColorChecker colour chart for image capturing with a Cannon EOS 600D. The images were standardized according to the colour chart and the ruler. They were then analysed by marking the bounds of the orange spots in ImageJ, which then calculated the area of orange colouration.
Scoring of behaviour
The quantification of male reproductive behaviour was done manually based on the obtained videos, using JWatcher v. 1.0 (Blumstein and Daniel 2007) as a quantification tool. The male behaviours that were scored have been described in detail previously by Liley (1966) and Houde (1997), and were as follows: frequency of sneaking behaviour, courtship display duration and frequency of gonopodial swing. A male sneak copulation attempt was recorded in situations where the male oriented himself behind the female, and then tried to thrust his gonopodium into the female’s gonopore to achieve a forced copulation. A courtship display was recorded when a male oriented himself in front of the female and bent his body in an “S”- shape while rapidly quivering his body. Gonopodial swings were defined as a forward bending of the gonopodium that was not accompanied by a sneak copulation attempt.
We included gonopodial swing frequency in the analysis, although there is some uncertainty regarding its function. Frequency of gonopodial swings seems to be surprisingly uncorrelated with general reproductive behaviour, and males typically perform the behaviour even when no females are around (Liley 1966, Roubertoux 1992, Gasparini 2009). It should be mentioned however, that some correlation with reproductive behaviour has been found; males that encounter receptive females perform gonopodial swings more often for example (Liley 1966). To increase explanatory power of the analysis, gonopodial swings were scored and included in the statistical analysis.
The total time the male spent following the female (defined as being within the total length of an average female away from the female) was also analysed manually using the videos recorded from the top view. The average duration of each following sequence was also scored.
Statistical analysis
Since we collected multiple types of behavioural data we used a principal component analysis (PCA) to collapse the variables into two components in order to reduce the number of inferential tests. The PCA included sneak copulation attempt frequency, total courtship duration, average duration of the courtship sequences, gonopodial swing frequency, the total following duration and the average duration of the following sequences. An analysis of the
Pearson correlation coefficients for the behaviours revealed several significantly correlated behaviours, which justified using a PCA. Since it is an assumption of the PCA, though not a critical assumption, that the variables are normally distributed, some of them had to be log transformed. The total time following, the average duration of the following sequences and the sneak copulation attempt frequency were log transformed before performing the PCA.
Only principal component one (PC1) and principal component two (PC2) had eigenvalues over 1, and thus were the only two components used in the model (Jackson 1993). PC1 and PC2 cumulatively explained 71 % of the variance. PC1 and PC2 were then used as separate response variables in two different general linear mixed models. In the general linear mixed models, we used brain size selection line as fixed factor and replicate as random factor. Male size, female size, area of orange colouration (corrected for body size) and time of day were used as covariates in the model. The models were ranked based on the Akaike information criterion (AIC), and the model with the lowest AIC value was chosen, since it explained the largest amount of variance in relation to the number of explanatory variables (Zuur et al.
2009). A total of 10 individuals were removed from the statistical analysis of both PC1 and PC2. Out of these, 1 individual was removed because he had a damaged gonopodium, 1 individual was removed because the female died before size measurement, and another individual was disregarded due to female aggression. The remaining 7 individuals were removed due to a lack of reproductive behaviour. Lack of reproductive behaviour is problematic when performing a PCA, because the PCA cannot handle zeroes in the data set.
Out of these 10 removed individuals, 5 were from the large brain size selection line and 5 were from the small brain size selection line leaving us with 25 individuals from each of the two groups. The difference in the area of orange colouration between the two brain size selection lines was analysed using a general linear mixed model. It included brain size selection line as fixed effect, replicate as random effect and log transformed, body size- corrected orange colouration as response variable. No individuals were removed from the analysis of the differences in area of orange colouration. The test statistics and the significance levels from the general linear mixed models for PC1, PC2 and the area of orange colouration were all generated by Type II Wald chi-square tests.
The principal component analysis and the general linear mixed models were performed using R 3.0.2 (R Core Team 2013). In R, the function princomp in the package stats was used for the PCA, and the function lmer in the package lme4 was used for the general linear mixed models. To retrieve the test statistics and significance levels from the general linear mixed models, the Anova function in the car package in R was used. Model diagnostics were performed for all models and residuals were roughly normally distributed, without signs of heteroscedasticity.
Results
Table 1. Loadings of the reproductive behaviours performed by male guppies that were analysed in the principal component analysis. Only components with higher eigenvalues than 1 (PC1 and PC2) are included.
Behaviour Loading PC1 Loading PC2
Gonopodial swing frequency
-0.076 -0.55
Sneak copulation attempt frequency
0.54 -0.04
Total courtship duration -0.33 -0.44
Average duration of courtship sequences
-0.20 -0.59
Total following duration 0.56 -0.19
Average duration of following sequences
0.50 -0.35
Figure 2. Matrix of Pearson correlation coefficients, displaying significant associations between behaviours and male and female traits for all male-female pairs. The circles are colour coded for direction of correlation (see colour scale), and the size of the circle indicates the strength of the correlation. The p-values are given inside each circle.
In the separate analysis of PC1 in relation to brain size selection line, the best fitting model according to the AIC values yielded no significant effect of brain size selection (χ2=0.665, df=1, p=0.415), female size (χ2=0.173, df=1, p=0.678) or male size (χ2=2.210, df=1, p=0.137). Figure 3 illustrates the mean difference in the fitted values of PC1 and PC2 for pairings with large- and small-brained males. However, the area of orange colouration had a significant negative effect on PC1 (χ2= 5.066, df=1, p=0.024) (fig. 4). Hence, males with more orange colouration performed less sneak copulation attempts, and followed the female during a shorter fraction of the time in the trials. In the analysis of PC2, the best fitting model generated no significant effects of brain size selection (χ2=0.804, df=1, p=0.370) (fig. 3), female size (χ2=0.192, df=1, p=0.662), male size (χ2=0.041, df=1, p=0.840) or area of orange colouration (χ2=0.215, df=1, p=0.643) (fig. 4). None of the interactions were significant for either PC1 or PC2.
For both the analysis of PC1 and the analysis of PC2, the best fitting model was large and included several interactions. In both analyses, a four-way interaction between brain size selection line, female size, male size and area of orange colouration was included. Even though this interaction was non-significant for both PC1 (χ2=0.154, df=1, p=0.695) and PC2 (χ2=0.609, df=1, p=0.435), removing it from the model resulted in a distinct increase of the AIC value. For PC1 the best-fit model had an AIC value of 155, while the next best model had an AIC value of 163. For PC2 the best-fit model had an AIC value of 141, while the next best model had an AIC value of 149.
Figure 3. Estimated effect of brain size selection on PC1 and PC2 from the general linear mixed model. Dots are
Figure 4. Estimated relationship between the area of orange colouration (corrected for body size) and PC1 and PC2 from the general linear mixed model. Lines represent estimated slope; shaded area represent 95 % confidence intervals. The rug in the bottom of each graph represents the fitted x-value value for each individual.
Figure 5. Depiction of the area of orange colouration (corrected for body size) for each replicate-brain size combination. The ends of the dotted red lines represent means for each replicate-brain size combination.
A general linear mixed model comparing the area of orange colouration between the brain size selection lines in the experiment revealed no significant difference in the area of orange colouration between large- and small-brained males (χ2=0.807, df=1, p=0.370), although all small-brained replicate groups had a lower mean area of orange colouration (fig. 5). This result supports a previous analysis on these selection lines (with a much larger sample-size), which demonstrated a significant difference in the area of orange colouration between the large- and the small-brained males (Kotrschal et al. in review).
Discussion
The results of our study indicate that relative brain size is not associated with reproductive behaviour in male guppies, at least not in a way that could easily be elucidated based on this data. The statistical analysis yielded non-significant results for the effect of brain size selection. However, area of orange colouration, corrected for body size, appears to influence reproductive behaviour.
Brain size selection did not have a significant effect on either PC1 or PC2. The reason for the non-existing effect of brain size selection is hard to determine. There is a difference in tail length between males of the two brain size selection lines (Kotrschal et al. in review), where males of the large brain size selection line have longer tails than the males of the small brain size selection line. Overall body condition is not significantly different however (Kotrschal et al. in review), which includes measures of body length (excluding tail length) and body weight. As mentioned in the introduction, males with longer tails perform sneak copulation attempts more often and courtship display less often (Karino and Kobayashi 2005, Karino and Kamada 2009). Furthermore, the males of the large brain size selection line are endowed with longer gonopodia (Kotrschal et al. in review), which has been found to correlate positively with sneak copulation attempts and female preference (Reynolds et al. 1993, Brooks and Caithness 1995). Consequently, these two traits (tail length and gonopodium length), disregarding area of orange colouration, have predicted outcomes for male behaviour that are unidirectional. In other words: the males of the large brain size selection line have longer gonopodia and longer tails, and are therefore predicted to employ the sneak copulation tactic more often because of this. Area of orange colouration was controlled for in the statistical analysis, but there is a possibility that the effects of tail length and gonopodium length mask a true difference between the groups. More precise estimates of the effects of these traits in future studies would be helpful, and collecting measurements of tail length and gonopodium length on the individuals used in this experiment might also bring clarity to the issue. A potential problem though with including more measurements in the current statistical model is that the statistical power decreases with the number of explanatory variables included. A repeated experiment with a larger sample size will therefore most likely also aid in separating the effect of the many possible variables that might influence the association between relative
A possible complication in determining the independent effect of brain size selection on male reproductive behaviour is that we do not yet know the true correlation between relative brain size and area of orange colouration. Theoretically the correlation might be very high.
Incidentally, the difference in the area of orange colouration between the brain size selection lines is 11.8 % (Kotrschal et al. in review), almost identical to the mean relative brain size difference (11 %). The significant effect of orange colouration might be due to the fact that the individually measured colour is a better proxy for relative brain size than the grouping of the selection lines. If we would have had a direct measure of relative brain size for the individuals in this study, it is possible that brain size selection line also would have a significant effect, assuming a high correlation between relative brain size and colour.
However, if the correlation was very high, determining the independent effects of relative brain size and the area of orange colouration would be practically impossible. In a statistical sense, relative brain size and area of orange colouration would become the same variable.
There is however a problem with using a direct measure of relative brain size. Since a lot of the variation in relative brain size is due to the unknown factors that are not related to brain size selection, including a direct measure of relative brain size would partly transform the experimental study to a correlational study.
An alternative explanation to the results of this study is of course that mating behaviour is not affected by relative brain size, at least not in the guppy. It might be that the two mating tactics are in fact equally cognitively demanding. It is hard to objectively tell which one requires the most cognitive processing, although the study by Fraser et al. (2014) indicates that sneak copulation might be more cognitively demanding. It could also be that more complex behaviour does not demand increased brain tissue, but this seems intuitively unlikely since brain tissue is the prerequisite for any sort of behaviour. Furthermore many previous studies have found a positive link between relative brain size and complexity of behaviour (Jacobs 1996, Lefebvre et al. 2002, Reader and Laland 2002, Byrne and Corp 2004, Kotrschal et al.
2013). As mentioned previously, there might still be a small difference that could be evident with a drastically larger sample size, although the effect size would probably be low. The question then becomes whether any observed difference would be meaningful in biological terms. Also, we could not address if there were differences in the effectiveness of the different behaviours assessed here between the brain size selection lines, because the number of copulations were very low. Although there was no difference between the brain size selection lines regarding the different behaviours, the outcome in reproductive success could still be different. This will be investigated using longer video recordings and paternity analysis in upcoming experiments on the brain size selection lines.
The results of the current study differ from the findings by Fraser et al. (2014). They compared brain gene expression in male sailfin mollies from two different treatments: a focal male with multiple other males and females, and a focal male alone with females. They found that sneaking behaviour positively correlated with an up-regulation of genes involved in learning, neurotransmission and locomotory behaviour. This suggests that the sneaking strategy might be more cognitively demanding than courtship display. It should be mentioned
however, that there are a number of important differences between the study by Fraser et al.
(2014) and our study. Firstly, guppies do not seem to have as distinct male mating strategies as the sailfin molly. Perhaps the more hard-wired sneaking strategy in the sailfin molly presents a stronger link to cognition than is the case for variation in courtship display in the guppy. Secondly, Fraser et al. (2014) was not studying brain size. It could be that other aspects of brain morphology, for instance the relative size of separate brain structures, are more strongly linked to variation in courtship behaviour. In contrast to the study by Fraser et al. (2014), Kolm et al. (2009) was studying relative brain size in relation to choice of mating tactic in brown trout (Salmo trutta), and they found a difference in relative brain size between the two mating tactics in brown trout. Again, there are important differences that might lead to the observed difference in outcome between that study and ours. For instance, the two tactics employed by the brown trout are related to major differences in both overall body size and ecology, which is not the case in the guppy.
The association between mating behaviour and relative brain size remains equivocal also in other taxa. For instance, Madden (2001) found a positive relationship between whole brain size and bower complexity in bowerbirds, but Day et al. (2005) could not confirm this observation. Instead, Day et al. (2005) found a positive relationship between cerebellum size (a vertebrate brain sub-structure) and bower complexity. They hypothesized that the positive relationship between cerebellum size and bower complexity might be due to the demand for procedural motor learning when constructing bowers. The discrepancy in the effect of whole brain size on bower complexity between the two studies on bowerbirds is hard to interpret.
Day et al. (2005) used slightly different measures for brain size and body weight, although the reanalysis of the data with the methods from Madden (2001) gave unaltered results. The results from the studies on bowerbirds are thus inconclusive, and seem to neither strongly support nor reject the results of our study. What they do highlight though is the potential need for inclusion of brain substructure size in these types of analyses, since natural and sexual selection on certain behaviours might affect different brain substructures differently (Barton and Harvey 2000).
Area of orange colouration had a significant effect on PC1, which was mainly represented by the number of sneak copulation attempts and following (following usually occurs when the male is trying to sneak copulate, see fig. 2). It is worth mentioning, however, that the interpretation of the principal components that were used in this study involves some measure of subjectivity. But this is an unavoidable consequence of reducing the number of dimensions of the data, and the benefit of solving the fact that several of the behaviours were correlated seemed to outweigh the cost in terms of interpretability. The observed effect of area of orange colouration on PC1 cannot be explained by the fact that the brain size selection lines also differ in area of orange colouration. This is because brain size selection line is controlled for in the model when analysing the effect of area of orange colouration on PC1. Furthermore,
reason for the discrepancy is probably a difference in statistical power (n=60 vs. n=180). The fact that the brain size selection lines seem to differ in area of orange colouration in the previous study, motivated us to include area of orange colouration in our statistical model.
It is hard to determine what is causing the observed relationship between area of orange colouration and reproductive behaviour. Since we did not manipulate the orange colouration, the observed effect is just a correlation. One possible explanation is that the female’s reaction to the male’s colouration might influence the male’s behaviour. Previous studies have showed that females seem to prefer more colourful male guppies (Kodric-Brown 1985, Kodric-Brown 1989, Brooks and Caithness 1995, Pitcher et al. 2003), and that ornamentation in guppies is costly and probably serves as an honest signal of condition (Endler 1980, Nicoletto 1991, Godin and McDonough 2003). More colourful male guppies have also been found to behave differently: they spend more time near a female (where the female is separated from the male by a Plexiglas slide) than less colourful males in a diet-manipulation experiment, possibly due to differences in female response (Kodric-Brown 1989). We did not find any relationship between the area of orange colouration and PC2, which mainly described courtship display and gonopodial swings. One possible explanation for this might be that overall investment in reproductive behaviour differs between colourful and dull males, and that this difference is mainly expressed in a reduction of sneak copulation attempts. This potential difference in reproductive investment could be due to a difference in the efficiency of mate attraction.
More colourful males might be more effective at finding a female to mate with, and can allocate their resources to survival instead. On the other hand, since males adjust their intensity of mating behaviour according to the operational sex-ratio (Rodd and Sokolowski 1995), and all males were virgins, this seems unlikely. Differences might possibly persist even in these extreme cases with heavily skewed operational sex-ratio, although it would seem unwise for an individual male fish to be hesitant in such a situation, regardless of colouration.
There were only two confirmed copulations in the trials. Both of them involved a male from the small brain size selection line, but since the incidence was so low this was not included in the statistical analysis. The low number of copulations was also an expected outcome, since the females had mated before the trials.
It is also possible that the selection experiment lead to an uneven distribution of regional size increase, and determining potential regional size differences in the selection lines would consequently be very informative. Of particular interest for this question is the telencephalon, since it seems to have an important role in male reproductive behaviour in fish (Kyle and Peter 1982, Bruin 1983, Braithwaite 2005). Fish with a removed telencephalon seem to perform some, albeit reduced, reproductive behaviour (Schwagmeyer et al. 1977). A planned future experiment will investigate telencephalon size using CT-scan analysis to determine potential size difference of the telencephalon in the selection lines, and might bring clarity to this issue. Currently, very few studies exist on the link between relative brain size and reproductive behaviour, particularly at the within-species level (Boogert et al. 2011b). Hence,
we do not yet know how much of the variation in mating behaviour that is driven by differences in brain morphology in vertebrates.
Future analyses on the collected data will be performed to strengthen these results. For instance, latency times for behaviours will also be included in the statistical analysis. There might be benefits of adding latency for certain behaviours to the statistical analysis, for example if there is a difference in reaction to stress between the two brain size selection lines.
With a randomly distributed incidence of the behaviour, the latency should theoretically decrease exponentially. This distribution can be modelled with a permutation-resampling test, and then compared to the actual latencies. There is also another type of analysis available for the type of data collected in this study; a Markov chain Monte Carlo (MCMC) estimation could have been used. The MCMC method is a type of multivariate statistical test that can accommodate our multiple behavioural response variables as they are, without the need for a PCA prior to the analysis. Future analyses on the data might be conducted with an MCMC model in order to compare between the two alternative approaches and ensure that the results are robust to the statistical method of choice. In addition, we can also measure female behaviour from the recorded trials. This might allow us to control for the influence of the female’s response on the male’s behaviour, although interactions between female and male behaviour are hard to disentangle. In future analyses we can use the distribution of behaviour over time in the recorded videos for both females and males, to determine cause and effect in male behaviour.
For future studies, repeating the current experiment might be a good idea for a number of reasons, although possibly with slight modifications. Firstly, since several males did not express any form of reproductive behaviour during the 25-minute trials, future experiments might benefit from using longer acclimation periods. This will lead to a higher percentage of individuals that can be included in the statistical analysis and it will consequently enhance the statistical power. Secondly, a potential problem during this project was that the selected females were not in the same stage of gestation. There might have been females that were extracted from their home tanks immediately after giving birth, rendering them receptive during the experimental trials. To exclude the risk of females being or becoming receptive during the experiments, females should ideally be deprived of males, and then mated simultaneously under controlled conditions prior to the trials. It would also be interesting to investigate potential differences in the flexibility of mating behaviour. The accuracy with which the male evaluates the female’s receptivity, and consequently the appropriate use of mating tactic, might depend on general cognitive ability.
Conclusions
In this study we did not detect the hypothesized effect of brain size selection on reproductive
reproductive success in guppies are needed. These will hopefully bring clarity to the possible association between sexually selected traits and brain size.
Acknowledgement
I would like to thank my supervisor Niclas Kolm for the excellent support and feedback throughout my project. I also want to thank Alberto Corral López for all the help with planning, performing and analysing the experiments, and Wouter van der Bijl for his substantial contribution to the statistical analysis. Furthermore I would like to give a big thank you to Isobel Booksmythe for her thoughtful feedback on my written report.
References
Aiello LC, Wheeler P. 1995. The expensive-tissue hypothesis - the brain and the digestive- system in human and primate evolution. Current Anthropology 36: 199-221.
Andersson MB. 1994. Sexual selection. Princeton University Press, Princeton.
Aubin-Horth N, Landry CR, Letcher BH, Hofmann HA. 2005. Alternative life histories shape brain gene expression profiles in males of the same population. Proceedings of the Royal Society B: Biological Sciences 272: 1655-1662.
Barton RA, Harvey PH. 2000. Mosaic evolution of brain structure in mammals. Nature 405:
1055-1058.
Bierbach D, Makowicz AM, Schlupp I, Geupel H, Streit B, Plath M. 2013. Casanovas are liars: behavioral syndromes, sperm competition risk, and the evolution of deceptive male mating behavior in live-bearing fishes. F1000Research 2: 75.
Blumstein DT, Daniel JC. 2007. Quantifying behavior the JWatcher way. Sinauer Associates, Sunderland.
Boogert NJ, Anderson RC, Peters S, Searcy WA, Nowicki S. 2011a. Song repertoire size in male song sparrows correlates with detour reaching, but not with other cognitive measures. Animal Behaviour 81: 1209-1216.
Boogert NJ, Fawcett TW, Lefebvre L. 2011b. Mate choice for cognitive traits: a review of the evidence in nonhuman vertebrates. Behavioral Ecology 22: 447-459.
Boogert NJ, Giraldeau LA, Lefebvre L. 2008. Song complexity correlates with learning ability in zebra finch males. Animal Behaviour 76: 1735-1741.
Braithwaite VA. 2005. Cognitive ability in fish. Fish physiology 24: 1-37.
Brooks R, Caithness N. 1995. Female choice in a feral guppy population - are there multiple cues? Animal Behaviour 50: 301-307.
Brown C, Garwood MP, Williamson JE. 2012. It pays to cheat: tactical deception in a cephalopod social signalling system. Biology letters 8: 729-732.
Bruin JPC. 1983. Neural correlates of motivated behaviour in fish. In: Ewert JP, Capranica RR, Ingle DJ (eds). Advances in vertebrate neuroethology, pp. 969–995. Plenum Press, New York.
Burns JG, Rodd FH. 2008. Hastiness, brain size and predation regime affect the performance of wild guppies in a spatial memory task. Animal Behaviour 76: 911-922.
Byrne RW, Corp N. 2004. Neocortex size predicts deception rate in primates. Proceedings of the Royal Society B: Biological Sciences 271: 1693-1699.
Catchpole CK. 1996. Song and female choice: Good genes and big brains? Trends in Ecology
& Evolution 11: 358-360.
Chapman BB, Morrell LJ, Krause J. 2009. Plasticity in male courtship behaviour as a function of light intensity in guppies. Behavioral Ecology and Sociobiology 63: 1757-1763.
Chittka L, Niven J. 2009. Are Bigger Brains Better? Current Biology 19: R995-R1008.
Crockford C, Wittig RM, Seyfarth RM, Cheney DL. 2007. Baboons eavesdrop to deduce mating opportunities. Animal Behaviour 73: 885-890.
Darwin C. 1871. The descent of man, and selection in relation to sex. D. Appleton and company, New York.
Day LB, Westcott DA, Olster DH. 2005. Evolution of bower complexity and cerebellum size in bowerbirds. Brain Behavior and Evolution 66: 62-72.
Devoogd TJ, Krebs JR, Healy SD, Purvis A. 1993. Relations between song repertoire size and the volume of brain nuclei related to song - comparative evolutionary analyses
amongst oscine birds. Proceedings of the Royal Society B: Biological Sciences 254:
75-82.
Dunbar RIM. 1998. The social brain hypothesis. Evolutionary Anthropology 6: 178-190.
DuVal EH. 2007. Cooperative display and lekking behavior of the lance-tailed manakin (Chiroxiphia lanceolata). Auk 124: 1168-1185.
Endler JA. 1980. Natural selection on color patterns in Poecilia reticulata. Evolution 34: 76- 91.
Farr JA. 1980. Social behavior patterns as determinants of reproductive success in the guppy, Poecilia reticulata Peters (Pisces, Poeciliidae): An experimental study of the effects of intermale competition, female choice, and sexual selection. Behaviour 74: 38-91.
Fraser BA, Janowitz I, Thairu M, Travis J, Hughes KA. 2014. Phenotypic and genomic plasticity of alternative male reproductive tactics in sailfin mollies. Proceedings of the Royal Society B: Biological Sciences 281: 20132310.
Fuller JL. 1979. Fuller BWS lines: History and results. In: Hahn ME, Jensen C, Dudek BC (eds). Development and Evolution of Brain Size, pp. 190-204. Academic Press, New York.
Garamszegi LZ, Eens M, Erritzoe J, Moller AP. 2005. Sexually size dimorphic brains and song complexity in passerine birds. Behavioral Ecology 16: 335-345.
Gasparini C. (2009). Sexual conflict in the guppy, Poecilia reticulata. Ph.D. Thesis.
University of Padua: Italy.
Godin JGJ. 1995. Predation risk and alternative mating tactics in male Trinidadian guppies (Poecilia reticulata). Oecologia 103: 224-229.
Godin JGJ, McDonough HE. 2003. Predator preference for brightly colored males in the guppy: a viability cost for a sexually selected trait. Behavioral Ecology 14: 194-200.
Gross MR. 1996. Alternative reproductive strategies and tactics: Diversity within sexes.
Healy SD, Rowe C. 2007. A critique of comparative studies of brain size. Proceedings of the Royal Society B: Biological Sciences 274: 453-464.
Houde AE. 1997. Sex, color, and mate choice in guppies. Princeton University Press, Princeton.
Huber R, Rylander MK. 1992. Brain morphology and turbidity preference in notropis and related genera (Cyprinidae, Teleostei). Environmental Biology of Fishes 33: 153-165.
Isler K. 2011. Energetic trade-offs between brain size and offspring production: Marsupials confirm a general mammalian pattern. Bioessays 33: 173-179.
Iwaniuk AN, Hurd PL. 2005. The evolution of cerebrotypes in birds. Brain Behavior and Evolution 65: 215-230.
Jackson DA. 1993. Stopping rules in principal components analysis - a comparison of heuristic and statistical approaches. Ecology 74: 2204-2214.
Jacobs LF. 1996. Sexual selection and the brain. Trends in Ecology & Evolution 11: 82-86.
Jerison HJ. 1973. Evolution of the brain and intelligence. Academic Press, New York.
Karino K, Kamada N. 2009. Plasticity in courtship and sneaking behaviors depending on tail length in the male guppy, Poecilia reticulata. Ichthyological Research 56: 253-259.
Karino K, Kobayashi M. 2005. Male alternative mating behaviour depending on tail length of the guppy, Poecilia reticulata. Behaviour 142: 191-202.
Keagy J, Savard JF, Borgia G. 2009. Male satin bowerbird problem-solving ability predicts mating success. Animal Behaviour 78: 809-817.
Keagy J, Savard JF, Borgia G. 2012. Cognitive ability and the evolution of multiple behavioral display traits. Behavioral Ecology 23: 448-456.
Kiritome A, Sato A, Karino K. 2012. Influences of orange spot patterns and operational sex ratio on male mating behaviors in the guppy Poecilia reticulata. Ichthyological Research 59: 304-313.
Kodric-Brown A. 1985. Female preference and sexual selection for male coloration in the guppy (Poecilia reticulata). Behavioral Ecology and Sociobiology 17: 199-205.
Kodric-Brown A. 1989. Dietary carotenoids and male mating success in the guppy - an environmental component to female choice. Behavioral Ecology and Sociobiology 25:
393-401.
Kolm N, Gonzalez-Voyer A, Brelin D, Winberg S. 2009. Evidence for small scale variation in the vertebrate brain: mating strategy and sex affect brain size and structure in wild brown trout (Salmo trutta). Journal of Evolutionary Biology 22: 2524-2531.
Kotrschal A, Rasanen K, Kristjansson BK, Senn M, Kolm N. 2012. Extreme sexual brain size dimorphism in sticklebacks: A consequence of the cognitive challenges of sex and parenting? Plos One 7: e30055.
Kotrschal A, Rogell B, Bundsen A, Svensson B, Zajitschek S, Brannstrom I, Immler S, Maklakov AA, Kolm N. 2013. Artificial selection on relative brain size in the guppy reveals costs and benefits of evolving a larger brain. Current Biology 23: 168-171.
Kyle AL, Peter RE. 1982. Effects of forebrain lesions on spawning behavior in the male goldfish. Physiology & Behavior 28: 1103-1109.
Lefebvre L, Nicolakakis N, Boire D. 2002. Tools and brains in birds. Behaviour 139: 939- 973.
Liley NR. 1966. Ethological isolating mechanisms in four sympatric species of poeciliid fishes. Behaviour, Suppl. 13: 1-197.
Madden J. 2001. Sex, bowers and brains. Proceedings of the Royal Society B: Biological Sciences 268: 833-838.
Maguire EA, Gadian DG, Johnsrude IS, Good CD, Ashburner J, Frackowiak RSJ, Frith CD.
2000. Navigation-related structural change in the hippocampi of taxi drivers.
Procedings of the National Academy of Sciences of the United States of America 97:
4398-4403.
Magurran AE, Nowak MA. 1991. Another battle of the sexes - the consequences of sexual asymmetry in mating costs and predation risk in the guppy, Poecilia reticulata.
Proceedings of the Royal Society B: Biological Sciences 246: 31-38.
Magurran AE, Seghers BH. 1990. Risk sensitive courtship in the guppy (Poecilia reticulata).
Behaviour 112: 194-201.
Mink JW, Blumenschine RJ, Adams DB. 1981. Ratio of central nervous system to body metabolism in vertebrates - its constancy and functional basis. American Journal of Physiology 241: R203-R212.
Møller AP. 2010. Brain size, head size and behaviour of a passerine bird. Journal of Evolutionary Biology 23: 625-635.
Nicoletto PF. 1991. The relationship between male ornamentation and swimming
performance in the guppy, Poecilia reticulata. Behavioral Ecology and Sociobiology 28: 365-370.
Northcutt RG. 2002. Understanding vertebrate brain evolution. Integrative and Comparative Biology 42: 743-756.
Pawlowski B, Lowen CB, Dunbar RIM. 1998. Neocortex size, social skills and mating success in primates. Behaviour 135: 357-368.
Peters S, Searcy WA, Nowicki S. 2014. Developmental stress, song-learning, and cognition.
Integrative and Comparative Biology 54: E164-E164.
Pitcher TE, Neff BD, Rodd FH, Rowe L. 2003. Multiple mating and sequential mate choice in guppies: females trade up. Proceedings of the Royal Society B: Biological Sciences 270: 1623-1629.
Pitnick S, Jones KE, Wilkinson GS. 2006. Mating system and brain size in bats. Proceedings of the Royal Society B: Biological Sciences 273: 719-724.
Pollen AA, Dobberfuhl AP, Scace J, Igulu MM, Renn SCP, Shumway CA, Hofmann HA.
2007. Environmental complexity and social organization sculpt the brain in Lake Tanganyikan cichlid fish. Brain Behavior and Evolution 70: 21-39.
R Core Team. 2013. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org/.
Ratcliffe JM, Fenton MB, Shettleworth SJ. 2006. Behavioral flexibility positively correlated with relative brain volume in predatory bats. Brain Behavior and Evolution 67: 165- 176.
Reynolds JD, Gross MR, Coombs MJ. 1993. Environmental conditions and male morphology determine alternative mating behavior in Trinidadian guppies. Animal Behaviour 45:
145-152.
Rodd FH, Sokolowski MB. 1995. Complex origins of variation in the sexual-behavior of male trinidadian guppies, Poecilia reticulata - Interactions between social environment, heredity, body-size and age. Animal Behaviour 49: 1139-1159.
Roth G, Dicke U. 2005. Evolution of the brain and intelligence. Trends in Cognitive Sciences 9: 250-257.
Roubertoux PL. 1992. Courtship behavior in the male guppy (Poecilia reticulata); a genetic analysis. International journal of comparative psychology 5: 145-163.
Ryan MJ, Akre KL, Kirkpatrick M. 2009. Cognitive mate choice. In: Dukas R, Ratcliffe JM (eds). Cognitive ecology II, pp. 137-155. University of Chicago Press, Chicago.
Sawaguchi T, Kudo H. 1990. Neocortical development and social structure in primates.
Primates 31: 283-289.
Schillaci MA. 2006. Sexual selection and the evolution of brain size in primates. PLoS One 1:
e62.
Schneider CA, Rasband WS, Eliceiri KW. 2012. NIH Image to ImageJ: 25 years of image analysis. Nature Methods 9: 671-675.
Schwagmeyer P, Davis RE, Kassel J. 1977. Telencephalic lesions and behavior in the teleost Macropodus opercularis (L.): effects of telencephalon and olfactory bulb ablation on spawning and foamnest building. Behavioral biology 20: 463-470.
Sewall KB, Soha JA, Peters S, Nowicki S. 2013. Potential trade-off between vocal ornamentation and spatial ability in a songbird. Biology Letters 9: 20130344.
Sherry DF, Jacobs LF, Gaulin SJC. 1992. Spatial Memory and Adaptive Specialization of the Hippocampus. Trends in Neurosciences 15: 298-303.
Shettleworth SJ. 2010. Cognition, evolution, and behavior. 2nd ed. Oxford University Press, Oxford.
Sol D, Bacher S, Reader SM, Lefebvre L. 2008. Brain size predicts the success of mammal species introduced into novel environments. American Naturalist 172: S63-S71.
Sol D, Duncan RP, Blackburn TM, Cassey P, Lefebvre L. 2005. Big brains, enhanced cognition, and response of birds to novel environments. Proceedings of the National Academy of Sciences of the United States of America 102: 5460-5465.
Striedter GF. 2005. Principles of brain evolution. Sinauer Associates, Sunderland.
Wimer C, Prater L. 1966. Some behavioral differences in mice genetically selected for high and low brain weight. Psychological Reports 19: 675-681.
Zuur AF, Ieno EN, Walker N, Saveliev AA, Smith GM. 2009. Mixed effects models and extensions in ecology with R. Springer, New York.
