Smells Like Trouble

Introduction

Intraspecific aggression is commonplace across the animal kingdom, largely due to competition for limited resources, from food and shelter to territory and sexual partners (Stevenson & Rillich, 2016). Indeed, agonistic social interactions have a key role in reproductive success across many different taxa (Buena & Walker, 2008), including in crickets (Grylloidea) (Savage et al., 2005). However, fighting is metabolically costly (Hack, 1997) and, more seriously, often carries the risk of injury or death (Huntingford & Turner, 1987). As such, to minimise uneconomically favourable conflict that is likely not to be worth the potential resulting costs in fitness, escalation in aggression tends to be mediated, among a range of factors, by perceived resource value (Enquist & Leimar, 1987). Should an agonistic bout between two male crickets commence, escalation in aggression tends to stereotypically follow a relatively predictable probabilistic sequence, ranging from inexpensive, lower risk behaviours, such as antennal fencing, to riskier behaviours, climaxing in full grappling (Hack, 1997); these are illustrated in the ethogram used to categorise levels of aggression in this study, shown in Figure 2 (Stevenson et al., 2005).

Chemical communication is acknowledged to be key in the mediation of social interactions in insects (d’Ettorre & Hughes, 2008). While crickets appear to rely on acoustic and vibrational communication more so that chemical, it has been found that olfactory cues can be used by both field (Gryllus) and house (Acheta) species to determine the sex of another individual, with aggression in male crickets increasing in the presence of the odour of a female (Otte & Cade, 1976). However, the consensus on this is not conclusive, with a more recent study finding that, while the presence of conspecific females elicited increased agonistic behaviour, female olfactory cues alone did not have such an effect (Tachon et al., 1999). A 2008 study (Buena & Walker) noted that, as predicted by behavioural game theory models (Enquist & Leimar, 1987), resident males were increasingly aggressive, defending their territories more vigorously, when it was of a higher perceived value, which importantly was manipulated by varying the number of females’ chemical cues present. To further investigate the impact of female pheromonal cues on modulating male cricket aggression, we paired male Gryllus bimaculatus of similar sizes in a controlled arena under three different treatments (female pheromones, a single female, and a control with neither) and recorded various indicators of aggression. Filter paper that had been left in a female-only G. bimaculatus tank was used for the pheromonal treatment so, should our results align with Buena & Walker’s findings (2008) that aggression was correlated with number of females’ pheromones present (and therefore resource value), we would expect significantly increased aggression between individuals under this condition. However, should olfactory cues alone be insufficient to elicit an aggression response, as stated by Tachon et al. (1999), we would not observe such a trend, and instead the single female treatment, which provided the males with visual and potentially acoustic/vibrational cues (in addition to pheromonal cues of just one female), would result in the most significant increase in aggression.

Methods

Study animals

The G. bimaculatus for this study were sourced from a UK-based live feed supplier and separated into male and female isolation tanks upon arrival at the department. Commercial cricket food and water gel was provided ad libitum, and the crickets were euthanised via freezing at -20°C after completion of the study. As invertebrates, crickets are not under the regulation of the Animals (Scientific Procedures) Act 1986 however, the use of live organisms in this study was conducted in accordance with the ASAB Ethical Committee’s Guidelines for the ethical treatment of nonhuman animals in behavioural research (2023). Experiments were carried out between 4th and 11th March, 2025 and with the assistance of experienced technicians.

Pairing of males, arena design & fight procedure

Males were captured from the main tanks as required, selected for consistent and good body condition. Body size is acknowledged to be a major determiner of resource holding potential (RHP) in crickets (Dixon & Cade, 1986). To control for this, pronotum width was recorded for all individuals, and fighting pairs were approximately size-matched (see Appendix). The sufficiency of our size matching was verified through generalised linear modelling of size difference as a predictor for the aggression indicators that were found to be significant (see Results), and no significance was found for either (p=0.535 for Bouts.T~SizeDiff (negative binomial family with a log link function, used due to overdispersion), p=0.255 for MaxLevel.T~SizeDiff (Poisson family with a log link function)). It was ensured that paired males came from separate main tanks to avoid pre-established dominance, which would otherwise likely affect the dynamics of any future aggression (Stevenson & Rillich, 2012).

A set of custom hinged arenas was constructed from polypropylene containers (Figure 1), with final dimensions of 10W x 15L x 8H cm and a removable divider to facilitate a 2-minute acclimatisation period before each trial. Arenas were wiped with 70% ethanol and allowed to fully dry between uses to remove any persisting olfactory residues.

Trials lasted 10 minutes each (following the 2-minute acclimatisation period), and each was observed by two researchers for increased accuracy. The variables recorded for each trial are listed in Table 1. MaxLevel.1 and MaxLevel.T – the maximum level of aggression observed in the first agonistic interaction and overall, respectively – were determined using an ethogram of the stereotypical escalation sequence for agonistic interactions between male crickets (Stevenson et al., 2005) (Figure 2). Consistency between pairs of observers was verified through seven trial runs that were observed by all researchers, with interobserver reliability metrics calculated for each dependant variable (Table 2). Individual males were identified using distinguishing physical characteristics and non-toxic researcher-applied ink markings, and subsequently tracked using unique identifiers (see Appendix), to ensure that no individuals were re-fought soon after a previous fight and that no two same individuals were re-paired for a subsequent fight (to avoid pre-established dominance).

Treatments

Ten trials were run for each treatment. The three treatments were as follows: female pheromones (hereafter referred to as ‘Pheromones’) whereby a 10 x 15 cm cutting from a piece of filter paper left in the female main tank for 48 hours was placed on the arena floor; single female (‘Female’) whereby an individual female was placed in the arena, contained within a caged isolation tube (see Figure 1) that allowed the males to view her and detect any olfactory cues, but not physically access her (so that the aggression allocation of their time budgets could not be reduced by attempting copulation); and ‘Control’ with neither the filter paper nor single female present. The same female was used for all trials of the Female treatment so that varying female attractiveness could not interfere with male willingness to fight over the potential mate (male sexual preferences for cuticular hydrocarbons have been identified in T. oceanicus (Thomas & Simmons, 2010)).

Statistical analysis

Generalised linear models (GLMs) were used to assess the effect of treatment on each response variable. A Poisson family with a log link function was used for Bouts.T, MaxLevel.1, and MaxLevel.T, while a Gaussian family with an identity link function was used for Latency.1 and Duration.1. Model assumptions were verified by examining base R diagnostic plots, and no violations were observed. Overdispersion was assessed for Poisson models and not found. Post-hoc pairwise comparisons were conducted using Tukey’s HSD test. All statistical analyses were performed in RStudio version 2024.12.1+563.

Results & Analysis

Fight count

Fight count ranged from 0 to 9, with median (IQR) Bouts.T values for Female, Control and Pheromones treatments of 2.5 (2.5), 3.5 (3.0) and 6.5 (2.75) respectively. Our Poisson GLM (Table 3) revealed a significant effect of Treatment on the number of fights (x2 = 19.66, df = 2, p = 5.38x10-5), with an overdispersion ratio of 1.02. Post‐hoc Tukey tests (adjusted) indicated that the Pheromones group exhibited more fights than both the Female group (p < 0.001) and the Control group (p = 0.010), whereas there was no significant difference between Female and Control (p = 0.389), as shown in Figure 3, corresponding to an increase by a factor of approximately 2.63 (e0.9651) in the number of fights in the presence of multiple females’ pheromonal cues versus the presence of a lone female.

Maximum aggression level

Maximum aggression levels ranged from 0 to 6, with median (IQR) MaxLevel.T values for Female, Control and Pheromone treatments of 2.5 (1.75), 4.5 (3.0), and 5.0 (1.0) respectively. Our Poisson GLM (Table 4) revealed a significant effect of Treatment on maximum aggression (x2 = 8.54, df = 2, p = 0.014), with an overdispersion ratio of 0.82. Post‐hoc Tukey comparisons indicated that the Pheromones group reached significantly higher peak aggression than the Female group (p = 0.013), whereas neither Pheromones vs. Control (p = 0.288) nor Female vs. Control (p = 0.339) differed significantly, as shown in Figure 4. This means that peak aggression doubles (e0.6931 = 1.9999) in the presence of multiple females’ pheromonal cues versus the presence of a lone female.

Other indicators of aggression

No significant effects emerged for duration of (Duration.1) or time to (Latency.1) first bout in Gaussian GLMs (p > 0.05 for all coefficients), and maximum escalation level of first bout (MaxLevel.1) showed no effect of Treatment in a Poisson GLM (p > 0.05 for all coefficients) (Table 6). Descriptive statistics for these three variables are shown in Table 5.

Discussion

Our findings that pheromonal cues from multiple females increases male aggression compared to the physical presence of a single female, namely a 2.63 factor increase in the number of agonistic bouts with double the maximum intensity, appears consistent with those of Buena & Walker (2008) in that cues representing a greater number of potential mates is indeed seemingly associated with a higher perceived resource value. This was demonstrated via increased aggression, as predicted by the behavioural game theory models of Enquist & Leimar (1987). It is possible that male crickets interpret multi-female pheromone signatures as a higher-reward environment, leading to greater motivation to establish dominance, manifesting as increased risk tolerance to engage in potentially more costly aggressive behaviours (Hack, 1997).

These results also serve to highlight the great importance of olfactory cues with respect to the mediation of intrasexual competitive behaviour in G. bimaculatus as identified by Otte & Cade (1976) but contrary, however, to the findings of Tachon et al. (1999) that olfactory cues alone were insufficient to increase male aggression. Indeed, in the decades following Tachon et al.’s publication, there has been mounting evidence of the ability of pheromones to elicit a range of different behaviours in crickets (Nagamoto, Aonuma & Hisada, 2005), although much of this seems to revolve around female pheromones eliciting mating behaviour responses in males and male pheromones eliciting agonistic responses in other males (as opposed to female olfactory cues eliciting agonistic competitive behaviour in males) (Iwasaki & Katagiri, 2008). As such, our study and its findings may be of particular significance.

Limitations & potential future research

It was unexpected that the Female treatment yielded no increased aggression (and that it in fact produced the lowest bout count and total intensity of all three treatments, although not significantly below Control). This is also at odds with the findings of Tachon et al. (1999) however, increased male aggression in the presence of conspecific females has been consistently observed across taxa (Smith et al., 1994; Kitchen, Cheney & Seyfarth, 2005; Petrusková et al., 2007), with the exception of sex-role-reversed female-dominated social structures such as that of the spotted hyaena (Curren et al., 2015). Therefore, this instead likely points to two potential flaws in our experimental design.

Firstly, the presence of the isolation tube containing the female will have altered the spatial dynamics of the arena – for example, it was observed on multiple occasions that the apparently less dominant of the two males would hide behind the tube, thus reducing the likelihood of physical encounters with the other male. A future improved study might employ the use of an external audience zone for the female to reside in, where she can be observed by the males without physical intrusion of the arena space.

Secondly, while we used one particular female in an attempt to maintain consistency and limit variability in female attractiveness confounding the aggression responses from one trial to the next (Thomas & Simmons, 2010), this may have in fact biased the results of the Female treatment group due to a failure to control for attractiveness characteristics of this female. For example, it has been f0und that the quality and quantity of female crickets’ cuticular pheromones is highly age dependant, with courtship behaviour more likely in males when exposed to cues of younger females (Nagamoto, Aonuma & Hisada, 2005). This same study, however, observed that older, ‘less attractive’ females can in fact elicit aggressive behaviours in males. This relationship not only meant that the age characteristics of our chosen female will have likely had a significant impact on the observed male behavioural responses, but that this will not have been a simple case of a younger female leading to more competitive fighting and vice versa, since the relationship at play appears to be substantially more complex. Further experimentation with a series of age-controlled trials would be valuable in exploring and disentangling this phenomenon.