Female–female aggression in Bactrocera tryoni (Diptera: Tephritidae) and the influence of fruit quality on combat intensity

Frugivorous tephritid (Diptera: Tephritidae) females compete over access to fruit for oviposition through aggressive interactions. These aggressive displays are for oviposition site maintenance to reduce the probability of subsequent larval competition. While female aggressive behaviours have been described for several frugivorous tephritid species, studies quantifying behavioural frequencies and sequences and examining how quality of the host fruit might modify the intensity of aggressive behaviours are minimal or absent. We used behavioural analysis software of video playback to describe and quantify antagonistic behaviours between pairs of Bactrocera tryoni females and measured changes in the frequency of behaviours when females were defending three fruit types known to vary in their quality for offspring development. Seven behaviours were identified as part of competitive contests between B. tryoni females, which were not performed in any regular order or with any obvious escalation in the intensity of an aggressive display. Crabbing, [wing] supination and pushing were the most common behaviours, constituting 78% of all observed aggressive behaviours. Increasing fruit quality resulted in aggressive behaviours happening significantly sooner and more often. Our results are similar to previous studies in the types of behaviours exhibited by female frugivorous tephritids but are contrary to other studies in that no sequential pattern or escalation of behaviours was documented. Increased female investment in defence of higher quality hosts aligns with theoretical predictions but has not been previously tested.


INTRODUCTION
Interference competition is a frequent type of ecological interaction, where competing organisms directly influence others by seeking to limit or exclude access to a common resource (Yip 2014).Interference competition may lead to the displacement of one species from the range of a superior competing species (Peiman & Robinson 2010), or may be the mechanism by which the fundamental niche of the subordinate species contracts, affecting resource partitioning and long-term coexistence (Ziv et al. 1993).Within interference competition, aggression is commonly associated with resource defence (Benelli, Desneux, et al. 2015).A variety of aggressive behaviours can be employed, ranging from low-cost antagonistic displays to high-cost contact actions (Hsu et al. 2006).In general, individuals that engage in more intense bouts of aggression have a higher probability of gaining access to the contested resource and are thus considered to have a greater competitive ability (Syme 1974).
One group of organisms for which there has been extensive competition research are insects of the family Tephritidae-the true fruit flies (Benelli 2014a).Parental females of frugivorous tephritids predominantly oviposit into the flesh of host fruit, where the resultant larvae feed and develop (Ansari et al. 2012).Among tephritids, competition occurs in two major ways-through behavioural interference competition between adults and through exploitative competition for resources among larvae (Duyck et al. 2004).The resource is always the fruit, with its combined use as an adult oviposition site and a larval feeding site.Tephritid females employ two main behavioural mechanisms to maintain single oviposition sites, thus increasing the chances of their offspring developing in a non-competitive environment (Benelli 2014b).The first behaviour is host marking, in which an oviposition deterring pheromone (ODP) is deposited over the fruit surface after oviposition to deter future oviposition by conspecific females (Fletcher & Prokopy 1991).The second behavioural tool involves female-female contests for oviposition sites (Benelli, Donati, et al. 2015).The use of ODPs has been observed in at least 27 frugivorous tephritid species across multiple genera; however, the use of ODPs is not universal across the frugivorous tephritids (Benelli, Giunti, et al. 2014).Bactrocera, the focus of this paper, is one genus for which ODPs are not known, while instead inter-female aggression is more frequent and intense (Shelly 1999).
The adaptive function of territorial aggression in female Bactrocera is presumed to be associated with its flow-on effects for limiting larval competition (Benelli, Romano, et al. 2015).By defending fruits (even temporarily), females may bestow their larvae with a 'head start' in growth over unrelated larvae, providing a competitive advantage for host fruit resources.Female-female contests generally consist of aggressive displays that rely on wing waving, chasing, pouncing and 'boxing' (Benelli, Daane, et al. 2014).In a study of Bactrocera tryoni (Froggatt), it was reported that females on fruit, even those already engaged in oviposition, were easily agitated by intruders and made threat displays that intermittently intensified to head-butting and pushing (Pritchard 1969).These aggressive behaviours are very similar to those subsequently described for B. dorsalis (Hendel) (Shelly 1999) and B. oleae (Rossi) (Benelli 2014b).In all three species, females react rapidly to the presence of intruding individuals, halting oviposition to confront intraspecific females.During these aggressive interactions, body size appears to be a key determinant of fighting success, with larger females displaying competitive success (Duyck et al. 2004;Shelly 1999).Exhibiting similar behaviour, but for a different purpose, B. oleae may also use aggressive displays to gain preferential access to bacterial food sources occurring on olive leaves and fruit (Sacchetti et al. 2008).Pritchard's (1969) study of B. tryoni, while historically important in tephritid behavioural research, is nevertheless limited because of the tools available to him.To better understand the antagonistic events that occur among competing B. tryoni females, we carried out descriptive and quantitative observational experiments to more comprehensively describe the individual behaviours, to determine the sequence and frequency in which they occur, and to determine if female size influences the initiation or outcome of interactions.Further, and novel for Bactrocera competition research, we manipulated host fruit type to determine if varying host quality influences female B. tryoni competitive behaviours.Fruit quality, as assessed by its suitability for offspring development (Silva et al. 2020;Silva & Clarke 2021), directly influences female B. tryoni response time to and memory duration of fruit (faster and longer, respectively, as fruit quality improves) (Silva et al. 2020), the likelihood of using the fruit for oviposition (increasing as fruit quality improves) (Silva & Clarke 2020a), and the likelihood of using a fruit with an existing larval infestation for oviposition (high in good fruit, rare in poor fruit) (Silva & Clarke 2021).Given these consistent differences in female behaviour with changes in fruit quality, we hypothesized that females should become increasingly more aggressive and exhibit more antagonistic behaviours as the quality of host improves from poor to medium to good.In addition, it was hypothesized that larger females would win more bouts than smaller individuals during competitive interactions to obtain a fruit.

General approach
This study used behavioural observation to describe and quantify antagonistic interactions between Bactrocera tryoni females on fruit in small arenas.Building on previous descriptions of aggression among fruit flies of the genus Bactrocera (Benelli 2014b;Pritchard 1969;Shelly 1999), the study first identified and described each component behaviour within the antagonistic displays.Once individual behaviours were identified, and so able to be coded, video recordings of interactions between competing females were then analysed using behavioural analysis software (BORIS, Behavioural Observation Research Interactive Software [Friard & Gamba 2016]).BORIS analysis provided quantification of the sequencing and proportion of time spent on each component behaviour, and this was used to construct a behavioural ethogram.Additionally, host-fruit type was changed to determine if host quality influenced the frequency and/or duration of antagonistic behaviours.Finally, the length of the dm wing cell of all videoed flies was measured to determine if female size influenced the frequency and outcome of competitive interactions.

Insects
Bactrocera tryoni pupae were obtained from laboratory colonies maintained by the Queensland Department of Agriculture and Fisheries (QDAF), Ecosciences Precinct, Brisbane, Australia.Colonies used were less than 6 months from the wild and had been maintained on a carrot-based diet (Heather & Corcoran 1985).Emergent adults were held in a nylon mesh cage (30 cm Â 30 cm Â 30 cm), supplied with water, sugar and yeast hydrolysate.The cage was held in an insectary at 26 C, 65% RH, and a photoperiod of 14 L:10D.When used in experiments, flies were 14-18 days old and sexually mature.

Initial identification of behaviours
A nylon cage (W32.5 Â D32.5 Â H77.0 cm) was set up outdoors between 10 AM and 2 PM, the peak time for oviposition in B. tryoni (Ero et al. 2011).An apple, a known B. tryoni host (Hancock et al. 2000), was hung from the top of the cage.Two B. tryoni females were introduced into the cage and as interactions took place between the females, notes were taken of the behaviours observed.Observations were carried out until one female oviposited.Observations were repeated with new pairs or with increased numbers of individuals.These preliminary observational data identified seven female antagonistic behaviours (see Results section), which were used for scoring behavioural sequences when analysing video playbacks.

Behavioural recording and size measurements
To fully quantify aggressive behaviours, high-resolution videos were taken of interacting flies.To ensure full visibility of behaviours, the observation arena was a single fruit piece cut in half sitting on the bottom of a small (10 cm Â 10 cm) Perspex box.Two female flies, one marked with acrylic paint and one unmarked, were introduced into the arena, and all behaviours videoed for a 1-h period.Preliminary observations showed no effect of host marking.Three types of host fruit were used during recordings, representing poor (Cherry tomato, Lycopersicon lycopersicum), intermediate (Red Delicious apple, Malus sylvestris) and good (mango, Mangifera indica) quality hosts based on their suitability for offspring development (Silva et al. 2020).Fifteen replicates were completed for each fruit type, with new flies and fruit used for each replicate.
Following recording, the length of the wing dm cell for each fly was measured using ImageJ software of wing micrographs (see Newman et al. 2021, for a discussion on the use of the length of the dm cell as a surrogate size measure in B. tryoni); dm cell length measurements allowed female size to be considered as a factor in explaining competitive interactions.

Video analysis
Using the BORIS package (Friard & Gamba 2016), the behavioural frequencies, duration and percentage of each of the seven identified antagonistic behaviours were added to and described in the behaviour coding map.In addition to the antagonistic behaviours, behaviour codes for walking, grooming, probing, oviposition and retreating were added.Each behaviour in BORIS coding represents an event.Each event was set either as a point event, if it was an instantaneous event with no duration, or as a state event if it had duration.Once the coding pad was set up, each video was set to playback at half-speed to allow clear observation of the behaviours.Records were made of each behaviour that took place, the time they started and finished, and where they took place.This was repeated for both females present in the video.Once the video analysis was complete, the BORIS lag sequential analysis was employed to determine the total number of transitions between all possible pairs of behavioural events.These behavioural frequencies were then used to establish the sequential pattern of B. tryoni antagonistic behaviours, and an ethogram and time budget were constructed.
The ethogram generated produced behavioural sequences, which were not expected (see Results section).As researchers, we were concerned that the duration of observation (1 h) and the combining of data for both females in an arena may have generated misleading results.To test for this, we reran the analyses using data from one randomly selected female per pair, or reran time budgets for the data generated by 15 min and 30 min after commencement.All post hoc analyses generated transition pathways and behavioural frequency data essentially identical to the original analysis, which was thus retained.The post hoc analyses of the reduced data sets are provided in Figure S1 and Table S1.

Analysis of fruit effects
Differences in the mean number of antagonistic events, mean time to first antagonistic event, mean time to probing/oviposition and mean duration of oviposition for each fruit type were compared using one-way ANOVA with a Tukey post hoc test and an alpha set at 0.05.Before analysis, data were evaluated for normality using Levene's test and transformed if required.In addition to the influence of fruit type on antagonistic behaviours, the effect of female wing size on winning or losing an aggressive interaction was analysed using chi-square tests.The analysis of size on behaviour was done by reducing size to a binary state for each pair of flies (larger or smaller) and then running chi-square tests for (i) the summed display behaviours, (ii) summed contact behaviours (see Table 1 for the behaviours involved), (iii) for 'retreat' and (iv) oviposition.Expected probability of a bigger or smaller female carrying out the behaviour was 50:50.All analysis was done in JMP version 16.

Identification and description of individual behaviours
Seven antagonistic behaviours were identified and described for female B. tryoni (Table 1).The behaviours between females ranged from display only with no physical contact (supination, crabbing and tiptoe); to physical antagonism but no contact (chasing); to direct contact (pushing, butting and wing-strike).All of these behaviours occurred both on and off the fruit.Emergent from these behaviours was an eighth behaviour-'retreat'-which was demonstrated by competitively inferior females following antagonistic behaviours directed at them.

Quantification of behaviours and behavioural sequence
For individual flies, walking and grooming occupied the majority of time (43.4% and 41.7%, respectively), with probing and oviposition occupying a further 10.1% of the time budget.Thus, antagonistic behaviours occupied only 4.8% of the combined 45-h observation period (Table 2).For the seven antagonistic behaviours, plus the additional events of probing, oviposition, walking, grooming or retreat, 8453 behavioural events were recorded from the 90 female B. tryoni: The sequence and transition of behavioural events are provided as an ethogram in Figure 1.
Antagonistic behaviours never began immediately after females were released into the arena, but only after grooming or walking, behaviours were observed.Both behaviours could transition to any of the aggressive behaviours, and any of the behaviours (aggressive or otherwise) could transition back to grooming or walking.These behaviours are not shown in the ethogram to minimise the complexity of the diagram.Once antagonistic behaviours began, there was no specific sequence in which they then occurred, that is, there is no evidence that aggressive behaviours escalated from one to another.This is visually illustrated by the complex, 'web-like' appearance of the ethogram (Figure 1), rather than it appearing as a predominantly linear ethogram.Crabbing was the behaviour most frequently observed (711 occurrences), while wing-strike was the least common behaviour being observed only 10 times (Table 2).
Females already in the act of probing or oviposition would interrupt those core behaviours to antagonistically interact with the second female.A female interrupted from oviposition was almost equally like to engage in the behaviours of crabbing (27% of occasions), supination (31%) or pushing (23%) (Figure 1), reinforcing the apparently random nature of the individual behaviours displayed at any given time in the overall aggressive display.Differences between Figure 1 and Table 2 in the numbers of observed behaviours (e.g., probing, 406 observations in Figure 1, 418 observations in Table 2) are due to the ethogram not including transitions to grooming or walking.
F I G U R E 1 Ethogram of antagonistic behaviours of female Bactrocera tryoni competing for a single host fruit.The size of the behavioural boxes and number within represent the frequency in which the individual behaviours occurred that led to a transitional flow to another behaviour.The number associated with the arrow represents the proportion of transition frequencies made by the females to other behaviours from a given behaviour and will sum to 1.For example, from a total number of observed behavioural events recorded for crabbing, a proportion of 0.381 of all transitions led to further crabbing (after the first crabbing had finished), 0.237 led to supination, 0.130 led to pushing, 0.091 led to tiptoe, 0.083 led to probing, 0.033 led to butting, 0.032 led to retreat and 0.014 led to chasing.The ethogram is based on 45 replicate recordings, each 1 h long of two sexually mature females competing for access to a single fruit resource for oviposition.The ethogram thus represents the combined behaviours of 90 individuals.

Influence of host fruit quality on competitive interactions
Fruit type had a significant effect on the time taken for the first antagonistic interaction to occur between a pair of females (df = 2, F = 12.40, p < 0.001).Female flies took significantly longer to perform their first antagonistic behaviour when the fruit substrate was tomato (poor host) than they did for either apple (moderate host) or mango (good host), which were not significantly different from each other (Figure 2).For all antagonistic behaviours combined, significantly more occurred on mango than apple, and significantly more on apple than tomato (df = 2, F = 17.66, p < 0.001) (Figure 3).
At the individual behaviour level, fruit type significantly affected some, but not all antagonistic behaviours (Table 3).The frequency of crabbing was significantly higher on mango and apple than on tomato, but was not significantly different between mango and apple.The frequency of supination was significantly different between all three fruit types, being highest on mango, intermediate on apple and lowest on tomato.Significantly more chasing occurred on mango than tomato, with apple intermediate and not significantly different to either (Figure 4).Fruit type did not significantly affect the frequency of other behaviours (Table 3).

Role of size
Flies in our study had a mean dm cell length of 2.147 mm ± 0.0112, with a range of 1.932 to 2.322 mm.The size (smaller vs. larger) of females had no significant effect on the number of contact and display events that occurred (total display antagonistic events smaller: larger 265:275, χ 2 = 0.093, p = 0.761; total contact antagonistic events 130:134, χ 2 = 0.030, p = 0.862).Additionally, there was no significant difference between the number of F I G U R E 2 Mean (±SE) time (in seconds) taken for a female Bactrocera tryoni to perform the first antagonistic behavioural event against a conspecific female when on one of three host fruit types.N = 15 female pairs per fruit type.Columns surmounted by different letters are significantly different at p = 0.05.
F I G U R E 3 Mean (±1 SE) number of antagonistic events carried out by Bactrocera tryoni females against conspecific females when on one of three host fruit types.N = 15 female pairs per fruit type.Columns surmounted by different letters are significantly different at p = 0.05.
T A B L E 3 Results of one-way analysis of variance comparing the mean occurrence of Bactrocera tryoni female-female aggressive behaviours recorded on three fruit types (tomato, apple and mango).retreat or oviposition occurrences between larger and smaller females (total retreat events smaller: larger 52:64, χ 2 = 0.623, p = 0.430; total oviposition events 47:63, χ 2 = 1.170, p = 0.279).

DISCUSSION
Seven aggressive behaviours were identified for B. tryoni female-female interactions.Based on these behaviours a transition ethogram was created, which illustrated an extreme complexity of female-female competitive interactions in this species, something not previously documented in frugivorous tephritids.In line with theoretical expectations, host quality was shown to have an influence on female aggression with increasing intensity on higher quality fruit.However, contrary to theoretical expectation and prior studies, the size of females did not have any influence on the intensity of aggression, nor measures of 'winning' (number of oviposition events achieved) or 'losing' (number of retreating events).Each of these three key areas (i.e., aggressive behaviours and their transition, host quality and female size) is discussed individually below.However, before beginning this discussion, we acknowledge that the small size of the arena in which the videos were recorded may have artificially increased aggression and oviposition interruption, as competing females were always in proximity.This was unavoidable as a small arena was necessary to keep the subjects in focus and capture all behaviours, and without the video, the detailed quantification of behaviours and transitions could not be done.Beyond potentially changing the frequency of behaviours, we saw nothing between our preliminary visual observations in the larger arena and the small arena video data, which led us to believe that the arena size changed other aspects of aggressive behaviour.Nevertheless, the caveat of small arena size should be kept in mind.

Antagonistic behaviours
The seven behaviours identified for B. tryoni during this study are common behaviours recorded for a number of tephritid species.Examples of tephritids that employ similar antagonistic displays are Anastrepha ludens (Robacker & Hart 1985), B. oleae (Benelli 2014b), B. dorsalis (Shelly 1999), Ceratitis capitata (Briceño et al. 1999), Rhagoletis pomonella (Biggs 1972) and R. indifferens (AliNiazee 1974).Adding B. tryoni to this species list reinforces the point that the aggressive behaviours exhibited by frugivorous tephritids appears to be highly conserved across the family.While antagonistic behavioural events are common across many tephritid species, there appears to be differences in how they are exhibited.Based on behavioural observations, specific sequences of behavioural have been reported for B. oleae (Benelli 2014b).During the study, Benelli (2014b) placed 30 females into a Plexiglass ® testing arena (diameter: 40 cm, length: 25 cm) containing foliage and a stem of a living olive plant (Olea europea).The aggressive interactions on the stem and fruit were noted by an observer for 20 min or until the end of the aggressive interaction, while aggressive behaviours displayed in other areas of the testing arena were discarded.The data obtained from these recording were then analysed to determine the sequence of female aggressive behaviours.However, we observed no sequential pattern of aggressive behaviours by B. tryoni females, but rather a complex web where behaviours appeared to occur randomly with respect to each other.The small size of the arena may have created an environment where behavioural patterns were disrupted (e.g., if a less competitive female could not escape).However, in a parallel study conducted in a much larger arena with multiple fruit and resting sites, we have similarly observed no sequential behaviours or evidence of aggressive escalation (authors' unpublished data), and so we believe that the behavioural patterns observed in the small arena accurately reflect the real pattern.

Fruit quality
During this study, females became more aggressive and territorial over better quality fruit.The time taken for the first antagonistic event to occur was longest on the poor quality host (tomato), while the number of antagonistic interactions significantly increased as fruit quality increased, with more than double the number of events occurring on mango in comparison with tomato.These experimental observations are in line with the theory that predicts that females should spend more time defending high-quality resources (Díaz-Fleischer & Aluja 2003).In a previous study examining the influence of host quality (clean vs. fruit covered with a host marking pheromone [HMP], as well as fruit size) on female fighting, Díaz-Fleischer & Aluja (2003) noted that the number of female bouts increased with increasing host quality.The number of female fights was greater on clean fruit than on fruit marked with HMP, and females fought more on smalland medium-sized fruit than on larger hosts.The influence of fruit size on female aggression did, however, vary with social context.
We note that cherry tomato is a much smaller fruit than the other two fruits used, which may be an alternate reason why it was less defended.However, in a study of cherry fruit fly, AliNiazee (1974) found that females were more aggressive on smaller fruit in order to maintain a single oviposition site.Thus, we consider it unlikely that cherry tomato was less defended because of its size but rather because of its lower host quality.If so, this study extends earlier work with B. tryoni that shows that females of this highly polyphagous species can consistently rank host species based on differences in host location and how it is utilised (Silva et al. 2020;Silva & Clarke 2020a, 2021), a capacity exhibited by some polyphagous herbivores, which remains challenging for theory to explain (Silva & Clarke 2020b).

Female size
We found that the size of the female did not have an influence on their level of aggression nor the outcome of key events.These results were unexpected as adult body size is often considered as a potential predictor of the success of competing flies (Duyck et al. 2004), as this trait influences competitive outcomes in numerous tephritid species (Benelli 2014a;Dodson 1986;Duyck et al. 2004;Shelly 1999Shelly , 2000)).In line with theoretical expectations, Shelly (1999) demonstrated that the body size of B. dorsalis females is a key determinant of fighting success, with larger females winning 85% of bouts.During this study, continuous behavioural observations of wild fly combats were carried out, followed by the collection to participants for body size measurements.The index of body size was based on measurements of the lengths of the posterior edge of discal cells; however, no size range/variation was noted.The lack of size effect in our study may be due to a low variance in female size of our laboratory reared flies in comparison with wild flies.Wild B. tryoni collected from various fruit types had an dm cell range of approximately 1.5 to 2.30 mm (Newman et al. 2021), that is, larger flies could be more 50% bigger than the smaller flies.The size of females in our study fell within this range, but the size variation between females was much smaller, at only around 10%-15% size difference between the smallest and the largest.Due to this consistency between individuals, we hypothesise that there was simply not enough size difference between competing females to induce a significant size effect.Trials using flies of greater size variation would be necessary to confirm this effect.

Concluding remarks
While it is known that maintaining a single oviposition site is a trigger for female aggression in tephritids, the environmental factors that modify the intensity of aggression (Herczeg et al. 2016) are still not well understood.External modifiers could influence the collective aggressive displays or individual behaviours within the display.During this study, it was identified that the level of aggression, as well as three individual behaviours, differs with hosts of varying quality.To further add to this and fully understand the influence of the environment as a whole, additional research investigating other aspects of the environment is necessary.Here, it would be valuable to look at the role of resource density and competitor density on modifying aggression and how this might change in the presence of another trophic group.
The time budget (in seconds) of the behavioural events for Bactrocera tryoni females observed during competitive interactions over a single host fruit.Description of female-female Bactrocera tryoni antagonistic behaviours when competing for access to a single host fruit for oviposition.
T A B L E 2Note: The data are accumulated from video-replay of 45 h of pair-wise female-female interactions.'Butting','wing-strike' and 'retreat' are instantaneous events without measurable duration.T A B L E 1