Behavioural plasticity may require energetically expensive sensory and neural adaptations to detect, process, and respond to social cues. These costs could lead to selection against behavioural plasticity and its eventual loss. We show that males from the behavioural plastic alternative reproductive tactic (ART) in the swordtail fish Xiphophorus multilineatus have relatively larger brains, in addition to a trade off with testes size, that is not detected in the males from the behaviourally fixed ART. Given these costs, we consider the hypothesis that plasticity in mating behaviours is maintained due to intralocus tactical conflict, where a shared genome can constrain one or both ARTs from evolving to their optima. When we reduced any potential for intralocus tactical conflict by removing the behaviourally fixed ART from long-term breeding mesocosms, the males from the behaviourally plastic ART were less plastic and had smaller brains as compared to their counterpart from control mesocosms (both male ARTs). We also detected evidence for a genetic correlation between the ARTs for behaviour, which is required for intralocus conflict. Our findings suggest that intralocus tactical conflict could be maintaining behavioural plasticity, in which case behavioural plasticity may not be adaptative in some cases.

ARTs in the swordtail fish Xiphophorus multilineatus

Males of the high-backed pygmy swordtail fish Xiphophorus multilineatus (Rauchenberger et al. 1990) are classified into two genetically influenced ARTs (Lampert et al. 2010): behaviorally fixed males, who are larger than their counterparts due to a later maturation and increased growth time-period; contrastingly, behaviorally plastic males mature earlier and thus remain small for their adult life (Zimmerer & Kallman 1989). Therefore, in addition to differences between the ARTs in behavioral plasticity, there is a bimodal distribution in body size, and various degrees of tactical dimorphism (significant differences in the mean expression of a trait between the ARTs) for additional traits with known associations with reproductive success (e.g., body shape). This relationship between morphological differences (body size and shape) between the ARTs and differences in use of mating behavior (Zimmerer & Kallman 1988; Liotta et al. 2019, 2021a) allows for an easy classification of ARTs even prior to performing behavioral tests.

Measuring Behavioural Plasticity

Behaviourally plastic ART males perform two types of mating behaviours (courtship display and sneak-chase) with sneak-chase being more likely to lead to an attempted copulation (Smith et al. 2015). The sneak-chase behaviour is always used when a larger male competitor is present, however when sneaker males are alone with a female, there is extensive variation in whether males continue to use sneak-chase behaviour, or switch to also using courtship behaviour (Smith et al. 2015). Therefore, we can measure behavioural plasticity by determining propensity to use sneak-chase behaviour in the context of being alone with a female. We calculated a male’s propensity to sneak-chase (PSC) as the total number of sneak-chases weighted by the total number of behaviours (total sneak-chases plus total courtships), with PSC=1 indicating low behavioural plasticity and PSC=0 indicating high behavioural plasticity. In addition, we calculated a “plasticity score,” a more intuitive score of plasticity, by subtracting the PSC from the proportion of sneak-chase behaviour used by the behaviourally plastic males in the context of a competitor (100% sneak-chase) and added a secondary Y-axis with this score to all the necessary graphs. A male that only used sneak-chase when alone with a female had plasticity score=0 while a male that used only courtship when alone with a female had a plasticity score=1.

All males were isolated in 2.5L tanks for a minimum of 10 days prior to being tested. We measured behaviours of males in the social context of being alone with a female (Fig. 1), by placing a male in a 75.7L tank along with one female that was separated from the focal male during the acclimation period. After the 10-min acclimation period, the female was released, and the test started when the male and female came within 1cm of each other, and the male began interacting with the female. All tests were filmed for 10 minutes using a GoPro Hero3 or GoPro Hero Session camera (GoPro Inc., San Mateo, California, USA) and conducted between 10.00 and 15.30 h, since X. multilineatus are diurnal and actively search for mates and food in the wild during this time range. We scored males for two behaviours: number of courtship displays (i.e., males slowly glide in front of or alongside of the female; one display counted as males showing one side of their body) and number of sneak-chases (male darts at the female, and swims along parallel as the female flees; one sneak-chase counted as a male’s continuous parallel swim next to female).

Morphology

Individuals were photographed in their tanks using a Canon Power Shot (Canon Inc., Tokyo, Japan), with a ruler taped to the front for calibration. A small hand net with tightened netting material created a flat surface that was used to hold the fish flat against the glass. Images were analysed using ImageJ (Schneider et al. 2012) for body size (standard length, SL, distance from snout to end of caudal peduncle). After behavioural testing and morphological measurements, fish were sacrificed using MS222 and either immediately dissected (fish from mesocosms) or preserved in 90% ethyl alcohol (F2 males). Testes and brains were removed by dissection and weighed using an Amscope dissection scope at a resolution of 5.1MP. Relative testes size and relative brain size were calculated by dividing tissue weight by SL. We compared the relative brain sizes between the two methods for a separate sample of fish from the mixed mesocosm (fresh N=17, preserved 70% alcohol N=15) and found significant differences (see Supplemental Materials). Therefore, we only compared brain sizes between specimens that were weighed using the same method.

Costs of Behavioural Plasticity

We used F2 males from laboratory crosses to compare the brain and testis weights between the behaviourally fixed and behaviourally plastic ART males. Wild-caught males were mated with lab reared virgin females of known lineage (sire behaviorally plastic or behaviorally fixed ART). Females were then isolated into their own 2.5L tanks and allowed to drop fry. Their offspring were reared with 0-2 siblings up to sexual maturity after which time they were isolated into their own 2.5L tanks. Therefore, we controlled for any differences between the ARTs that could be due to mating experiences. We compared brain sizes for 9 males from the behaviourally fixed ART and 41 males from behaviourally plastic ART using a General Linear Model (LM) with brain weight as the dependent variable, ART as a fixed factor, and SL and relative testes weight as covariates.

Reducing IATC by removing behaviourally fixed ART males

We wanted to determine if removal of the behaviourally fixed ART from laboratory breeding mesocosms (i.e., reduction of IATC if present) would result in the behaviourally plastic ART males evolving towards being behaviourally fixed. We were also interested in the direction of change (reduction in use of either sneak-chase or courtship). Finally, if differences in behavioural plasticity were detected, we wanted to determine if there had also been changes in brain size, which would lend additional support to the hypothesis that larger brain size is a cost of plasticity, and that this cost could lead to its evolutionary loss if IATC was reduced.

Behaviourally plastic ART males from three different sources were scored for their behavioural plasticity. Breeding mesocosms were set up in 454L tanks using approximately 12 males and 12 virgin females. One mesocosm was “mixed,” starting with 50% of the behaviorally fixed ART males and 50% of the behaviorally plastic ART males. Males were sampled from this mixed mesocosm at two different times, 18 months apart (mixed #1, n=15; mixed #2, n=20). Two additional mesocosms were set up with the “behaviorally plastic only” males (BPO #1, n=17; BPO #2, n=22). All mesocosms had been breeding for over four generations by the time the males were sampled. We also examined the behavioral plasticity of wild-caught males from the Rio Tambaque, San Luis Potosi MX (n=40).

To test for potential changes in behavioural plasticity with removal of an ART, we used a Generalized Linear Model (GLM) with binomial distribution and a log link function. The dependent variable represents the variation in the propensity to sneak-chase when alone with a female (PSC). In the model, PSC was the number of sneak-chases weighted by the total number of events occurring in the trial (i.e., sneak-chases+courtship displays=total number of behaviours) and the fixed factor was the source of the males (behaviourally plastic only mesocosms, mixed mesocosms with both behaviourally plastic and behaviourally fixed males, and wild-caught males).

We determined if behavioural plasticity was correlated with brain size within the behaviourally plastic males and if there had been a change in brain size in the mesocosms where the behaviourally fixed ART were removed by using a subset of males from the mixed #2 (n=16) and BPO # 2 (n=11) mesocosms. Males were dissected for brain weights after being tested for behavioural plasticity. We used the same GLM as before (see above), but this time the fixed factor only included the two types of mesocosms (mixed and behaviourally plastic only) as no brain data was collected for the wild-caught males. To analyse if there were differences in brain size between the mixed mesocosms and the one where the behaviourally fixed ART males were removed, we used a GLM with a gamma distribution and a log link function to compare relative brain sizes (dependent variable) between these two sources (fixed factor).

Genetic correlation for behavioural plasticity across the ARTs

To test for a genetic correlation across the ARTs in relation to use of mating behaviours, we tested the behavioural plasticity of F2 behaviourally plastic ART males. Given that males will have the same ART as their sires, and therefore phenotype (behaviourally plastic or behaviourally fixed), we considered the lineage of a male’s dam in relation to the ARTs so that we could compare males with a maternal grandfather that was behaviourally plastic to males with a maternal grandfather that was behaviourally fixed. We used a GLM to examine variation in behavioural scores (PSC) within the behaviourally plastic ART males, with dam’s lineage as a fixed factor. We also included relative brain weight and standard length (SL, as body size has been shown to be correlated with behavioural plasticity; Liotta et al. 2021b) as covariates within the model, as well as an interaction between SL and dam’s lineage.