Introduction
Mutualistic relationships between ants and honeydew-producing, phytophagous hemipterans represent a highly successful innovation in the evolutionary history of these insects and an ecological keystone in terrestrial habitats (Hölldobler and Wilson, Reference Hölldobler and Wilson1990; Parker and Kronauer, Reference Parker and Kronauer2021). Typically, they are based on the provision of food (the honeydew produced by the hemipterans) in exchange for protection and they are thus defined as trophobiotic (Hölldobler and Wilson, Reference Hölldobler and Wilson1990). Ants may defend their partners against pathogens (e.g. Queiroz and Oliveira Reference Queiroz and Oliveira2001; Nielsen et al., Reference Nielsen, Agrawal and Hajek2010), adverse climate (Giannetti et al., Reference Giannetti, Mandrioli, Schifani, Castracani, Spotti, Mori and Grasso2021) or arthropod natural enemies (e.g. Majerus et al., Reference Majerus, Sloggett, Godeau and Hemptinne2007; Dao et al., Reference Dao, Meats, Beattie and Spooner-Hart2014), and may even help them to colonise new plants (Collins and Leather, Reference Collins and Leather2002). Due to the high diversity of both ants and hemipterans involved in these relationships, several possibilities may co-exist, but little is still known about what exactly ants offer in exchange for honeydew in most cases (Schifani et al., Reference Schifani, Peri, Giannetti, Colazza and Grasso2023a).
Many honeydew-producing hemipterans, including aphids, scale insects and whiteflies, are serious economic pests, and their relationship with ants can have negative impacts on their management. In particular, the major concern is the ability of some ants to defend their partners against arthropod natural enemies such as predators and parasitoids which may interfere with biological control (Franco et al., Reference Franco, Suma, Da Silva, Blumberg and Mendel2004; Vanek and Potter, Reference Vanek and Potter2010; Navarrete et al., Reference Navarrete, McAuslane, Deyrup and Peña2013; Cocco et al., Reference Cocco, Pacheco da Silva, Benelli, Botton, Lucchi and Lentini2021). However, interactions between these insects are rather complex, and baseline data on the behaviour of most species are missing. Relationships between ants and hemipterans range from facultative to obligate mutualists (Depa et al., Reference Depa, Kaszyca-Taszakowska, Taszakowski and Kanturski2020), and not all ant species attack the predators and parasitoids of their partners (Schifani et al., Reference Schifani, Peri, Giannetti, Colazza and Grasso2023a). Furthermore, some predators and parasitoids have developed adaptations to circumvent or even take advantage of ant protection (Völkl, Reference Völkl1992; Völkl et al., Reference Völkl, Liepert, Birnbach, Hübner and Dettner1996; Schifani et al., Reference Schifani, Peri, Giannetti, Colazza and Grasso2023a).
To understand the ecology and evolution of these multitrophic networks, the potential negative role of ants as mutualists of pest hemipterans must be analysed without overlooking the other services that ants may provide to the plants they visit, ranging from suppression of plant pathogens (Offenberg and Damgaard, Reference Offenberg and Damgaard2019) to predation of non-mutualist phytophagous insects (Styrsky and Eubanks, Reference Styrsky and Eubanks2007). Some efficient predatory ants may also be strongly protective of their hemipteran partners (Styrsky and Eubanks, Reference Styrsky and Eubanks2007; Offenberg et al., Reference Offenberg, Nielsen and Damgaard2019), but little is still known about this behaviour for most European ants that may act as pest predators (Campolo et al., Reference Campolo, Palmeri, Malacrinò, Laudani, Castracani, Mori and Grasso2015; Schifani et al., Reference Schifani, Castracani, Giannetti, Spotti, Reggiani, Leonardi, Mori and Grasso2020).
Lady beetles are key predators of ant-mutualist hemipterans in natural environments and are frequently employed as biological control agents in agroecosystems (Obrycki and Kring, Reference Obrycki and Kring1998). We carried out a field experiment to study the behavioural interactions between five European ant species associated with the walnut aphid Panaphis juglandis and two lady beetle species. These lady beetles are important predators and biological control agents of the walnut aphid (Kök et al., Reference Kök, Aktaç, Özdemir and Kasap2018; Czechowski et al., Reference Czechowski, Trigos-Peral, Maák and Vepsäläinen2019; Gao et al., Reference Gao, Liu, Feng, Wang and Lu2020): the smaller and native Adalia bipuntata and the larger allochthonous Asian species Harmonia axyridis (Brown et al., Reference Brown, Adriaens, Bathon, Cuppen, Goldarazena, Hägg, Kenis, Klausnitzer, Kovář, Loomans, Majerus, Nedved, Pedersen, Rabitsch, Roy, Ternois, Zakharov and Roy2007; Li et al., Reference Li, Li, Lövei, Kring and Obrycki2021). Our experiment aimed at describing the behaviours of the associated ant species and comparing their differences in relationship with the protective ability they were able to grant to the aphids against the two lady beetle species.
Materials and methods
Data collection was carried out at the Parma University Campus, Italy (44.7684, 10.3140), from 6 to 17 June 2022 (and daily from 11:00 to 15:00), when P. juglandis was abundant. A total of 12 walnut trees (Juglans regia L.) colonised by P. juglandis were selected based on the presence of five different ant species associated: Camponotus piceus (Leach, 1825), Ca. vagus (Scopoli, 1763), Crematogaster scutellaris (Olivier, 1792), Dolichoderus quadripunctatus (Linnaeus, 1771) and Lasius emarginatus (Olivier, 1792). The identity of these ants was ascertained before the beginning of the experiment by sampling two workers per each species and each tree, identifying them using a ZEISS Stemi 508 stereoscopic microscope and the keys provided by Seifert (Reference Seifert2018). Adults of H. axyridis were collected in the field, while A. bipunctata adults were acquired from Bioplanet (Italy). Both species were temporarily reared in plastic cages under laboratory conditions (T: 25 ± 1°C, R.H.: 55 ± 10%) and fed with field-collected P. juglandis aphids.
In each experimental trial, we carefully introduced one adult lady beetle of either species to the dorsal surface of a leaf containing at least 15 P. juglandis aphids attended by ant workers. The introduction of the lady beetle was conducted slowly to avoid causing any reaction in the ants or aphids by creating vibrations on the plant – experimental trials in which the introduction itself accidentally caused visible vibrations and reactions by aphids or ants were aborted and another leaf was taken. In all cases, only one ant species was present on each leaf, and the leaves were located at a height of about 1.5 m from the ground. Once the lady beetle was introduced, we filmed the insects on the leaf until at least one of the following events occurred: (i) the lady beetle left the dorsal surface of the leaf; (ii) all aphids were either killed by the lady beetle or left the dorsal surface of the leaf; (iii) five minutes have passed since the introduction of the ladybug. Videos were analysed with the software Solomon Coder (solomon.andraspeter.com) to collect data on the interactions of ants, lady beetles and aphids.
We recorded three parameters as continuous variables:
➢ Ant workers on the leaf. The number of ant workers per leaf at the beginning of each experimental trial.
➢ Lady beetle on the leaf (time). The cumulative time (seconds) during which the lady beetle stayed on the leaf since the first encounter with ants.
➢ Lady beetle with the aphids (time). The cumulative time (seconds) during which the lady beetle stayed in proximity (<3 mm) of the aphids.
Moreover, ten behaviours were recorded as binary variables (presence/absence):
➢ Lady beetle biting aphids. The lady beetle bites and/or eats at least one of the aphids.
➢ Aphids flee from the leaf. At least one of the aphids stops feeding and walks away. In all our observations, this behaviour was always involving multiple aphids simultaneously.
➢ Ants flee from the leaf. At least one of the ants abandons the leaf by walking away to other parts of the plant.
➢ Ants opening mandibles (threat). At least one of the ants opens its mandibles towards the ladybug, without a successful bite following; genuine but unsuccessful biting attempts are not distinguished from bite threats.
➢ Ants biting the ladybug. At least one of the ants successfully bites the ladybug, grasping any part of its body or appendages.
➢ Ants bending their gaster. At least one of the ants directs its gaster towards the lady beetle at a short distance. This behaviour corresponds to threatening the release or releasing toxic chemicals against an enemy, the two being normally indistinguishable in the field. There are slight differences in the movements performed by the examined species: in Camponotus spp. and L. emarginatus, the gaster passes under the body and the head; in Cr. scutellaris the gaster may be directed at almost any possible angle and the spatulate stinger connects with the target to release its venom topically; in D. quadripunctatus, the gaster is oriented laterally, diverging from the body axis.
➢ Ants chasing the ladybug. While the lady beetle is far from the aphids (>5 mm distance), at least one of the ants stops attending to the aphid colony to threaten or attack the lady beetle with one of the abovementioned behaviours (ants open mandibles (threat), ants biting the ladybug or ants bending their gaster).
➢ Ants falling off the leaf. At least one ant falls off the leaf after approaching the ladybug. Ants were never observed falling off the leaf under any other circumstances.
➢ Ants grooming. At least one ant cleans its mouthparts after biting the ladybug.
➢ Ants rescuing aphids. After the lady beetle grabs an aphid with its mouth (see above lady beetle biting aphids), at least one ant starts pulling the aphid in the opposite direction. We recorded whether the aphid was freed or not from the lady beetle because of this behaviour.
We conducted 12 experimental trials for each combination of ant and lady beetle species for a total of 120 trials (12 replicates × 2 lady beetle species × 5 ant species). Data were analysed using the software R and RStudio (R Core Team, 2022; RStudio Team, 2022). We preliminarily explored the data through generalised linear models considering the role of the lady beetle species (factor with two levels), the ant species (factor with five levels) and their interaction on each of the continuous or binomial variables recorded. The interaction term was never statistically significant, which led us to analyse the role of the ant or lady beetle species through separate statistical tests. For continuous variables (ant workers on the leaf, lady beetle on the leaf and lady beetle with the aphids), we analysed differences between ant or lady beetle species using Kruskal–Wallis tests, which were followed by Conover's tests of multiple comparisons from the PMCMRplus R package whenever significant differences were detected. The Bonferroni correction was applied to Conover multiple comparison tests. Binomial variables (lady beetle biting aphids, aphids flee from the leaf, ants flee from the leaf, ants opening mandibles (threat), ants biting the ladybug, ants bending their gaster, ants chasing the ladybug, ants falling off the leaf, ants grooming, ants rescuing aphids) were analysed by using χ 2 tests, which were followed by the analysis of standardised residuals if significant differences were detected.
Results
Concerning the continuous variables, there were no statistical differences between experiments conducted with the two lady beetle species (ant workers per leaf: H (1) = 0.06, P = 0.80; lady beetle on the leaf (time): H (1) = 3.13, P = 0.07; lady beetle near aphids (time): H (1) = 1.33, P = 0.25), while differences between ant species were always significant (ant workers per leaf: H (4) = 101.25, P < 0.001; lady beetle on the leaf (time): H (4) = 17.80, P = 0.001; lady beetle near aphids (time): H (4) = 25.50, P < 0.001). Regarding ant workers per leaf, pairwise comparisons revealed that all ant species differed significantly from each other (0.014 < P < 0.001), except for Ca. piceus and Ca. vagus which shared the lowest numbers of workers (P = 0.998), while D. quadripunctatus had the highest one (fig. 1A). The lady beetle on the leaf (time) was significantly different between D. quadripunctatus and Ca. piceus (P = 0.021), D. quadripunctatus and Cr. scutellaris (P = 0.05) and between Cr. scutellaris and L. emarginatus (P = 0.04) (fig. 1B). The lady beetle near aphids (time) differed between Ca. piceus and D. quadripunctatus (P < 0.001) and Ca. piceus and L. emarginatus (P = 0.005) (fig. 1C).
Behavioural data collected during the experiments, divided according to the identity of the ant species involved in the trials. Concerning continuous numerical data (A–C), groups marked with the same lowercase letter were not significantly different according to pairwise comparisons. In the case of binomial data (D–L), groups significantly different than the expected are highlighted using asterisks according to the significance level (*P ≤ 0.05; **P ≤ 0.005; ***P ≤ 0.001).
For any of the other recorded binary variables, there were no statistically significant differences between experiments conducted with the two lady beetle species, while all of them differed significantly between trials conducted with different ant species except for the ants opening mandibles (threat) behaviour (table 1). The analysis of standardised residuals of χ 2 tests revealed that lady beetle biting aphids, aphids flee from the leaf and ants flee from the leaf behaviours were all observed more frequently than the expected with Ca. piceus (P < 0.001, fig. 1D–F). The frequency of ants bending their gaster was lower than expected in D. quadripunctatus (P = 0.002, fig. 1I). Ants chasing the lady beetle were observed less frequently than expected in Ca. piceus, and more frequently than the expected in D. quadripunctatus and L. emarginatus (P < 0.001; fig. 1J). Finally, ants grooming was observed more frequently than expected in Ca. vagus (P < 0.001; fig. 1L).
Results of χ 2 analyses of differences between lady beetles and among ant species for nine binomial behavioural variables recorded in this study
Significant differences are highlighted in bold.
The ants rescuing aphids behaviour was the only one not to be statistically analysed as it was exhibited only twice (fig. 2). In both cases, this behaviour was performed by Cr. scutellaris, once interacting with A. bipunctata, and once with H. axyridis. In the first case, the ant successfully managed to free the aphid from the ladybug, while in the second case, both the lady beetle and the ant kept pulling the aphid in opposite directions beyond the time duration of the video.
Aphid rescue behaviour performed by Cr. scutellaris in filmed interactions with A. bipunctata (left) and H. axyridis (right). In both cases, an ant worker is shown attempting to save a P. juglandis aphid grabbed by the ladybug.
Discussion
The high diversity of interactions between ants, their hemipteran partners and the arthropod natural enemies of the hemipterans is still undocumented in most cases, but of high interest in evolutionary terms as well as in an applied perspective for pest management and biological control (Oliver et al., Reference Oliver, Leather and Cook2008; Depa et al., Reference Depa, Kaszyca-Taszakowska, Taszakowski and Kanturski2020; Parker and Kronauer, Reference Parker and Kronauer2021; Castracani et al., Reference Castracani, Giannetti, Spotti, Schifani, Ghizzoni, Delaiti, Penner, Leonardi, Mori, Ioriatti and Grasso2023; Schifani et al., Reference Schifani, Peri, Giannetti, Colazza and Grasso2023a). Our results highlight how a large proportion of this variation may depend on the identity of the ant species (Völkl et al., Reference Völkl, Liepert, Birnbach, Hübner and Dettner1996; Schifani et al., Reference Schifani, Peri, Giannetti, Colazza and Grasso2023a).
On one hand, D. quadripunctatus and L. emarginatus were the most effective species in the protection of aphids. Workers of D. quadripunctatus were on average more numerous than those of L. emarginatus and used fewer chemical attacks, but the behaviour of both species similarly caused the aphids to stay safe. Both aggressively attacked the lady beetles even if they were not close to the aphids, causing them to abandon the leaves rapidly. These results suggest that D. quadripunctatus can be a pugnacious species (Schifani et al., Reference Schifani, Giannetti, Csősz, Castellucci, Luchetti, Castracani, Mori and Grasso2022), dismissing the idea that it does not defend the associated P. juglandis colonies from lady beetles as stated by Czechowski et al. (Reference Czechowski, Trigos-Peral, Maák and Vepsäläinen2019).
On the other hand, the attacks of Ca. piceus were mostly ineffective at repelling the lady beetles, and its workers (which were normally few in numbers) rarely attacked the lady beetles if these were not close to the aphids, and often completely abandoned the leaf soon after the lady beetle attacks on the aphids started. This resulted in the lady beetles spending considerably more time on the leaves, and frequently killing aphids, often to the point of causing the whole aphid colony to flee.
An intermediate performance in terms of aphid protection was recorded for Ca. vagus and Cr. scutellaris. Ca. vagus workers occurred in small numbers like Ca. piceus but are much larger than any other ant species observed in this study (Seifert, Reference Seifert2018). Stronger biting force because of larger size may have determined their more effective protective service compared to Ca. piceus, and the higher frequency of self-cleaning behaviour after bites. Cr. scutellaris, an aggressive and dominant species of the canopy (Castracani et al., Reference Castracani, Bulgarini, Giannetti, Spotti, Maistrello, Mori and Grasso2017; Seifert, Reference Seifert2018; Giannetti et al., Reference Giannetti, Castracani, Spotti, Mori and Grasso2019, Reference Giannetti, Schifani, Gugliuzzo, Zappalà, Biondi and Grasso2022; Schifani et al., Reference Schifani, Giannetti, Csősz, Castellucci, Luchetti, Castracani, Mori and Grasso2022, Reference Schifani, Giannetti and Grasso2023b, Reference Schifani, Peri, Giannetti, Alınç, Colazza and Grasso2023c), was the only species to perform the aphid rescue behaviour. Considering that Cr. scutellaris queens sometimes store in special chambers of their nests living P. juglandis aphids during the earlier stages of colony foundation (Giannetti et al., Reference Giannetti, Mandrioli, Schifani, Castracani, Spotti, Mori and Grasso2021), our observations on the aphid rescue behaviour reinforce the idea of a special relationship between the two species. Both Ca. vagus and Cr. scutellaris often did not attack lady beetles that were not close to the aphids, and in their presence, lady beetles managed to kill some aphids. However, only in the case of Ca. vagus whole aphid colonies were observed to flee.
It was interesting to observe how ants had similar interactions with the two lady beetle species. Ant–lady beetle interactions can sometimes deeply differ based on the specific characteristic of the lady beetle species involved, with some specialised species being completely immune to ant attacks or even taking advantage of ant presence (Liere and Perfecto, Reference Liere and Perfecto2008). Apart from their geographic origin, A. bipunctata and H. axyridis are not known to possess any highly specific adaptation to cope with ants, but differ in size, with the latter being considerably larger than the former. However, the larger ants examined in our experiment, Ca. piceus and Ca. vagus, performed as good or worse than the remaining smaller species, suggesting that size is not a key determinant in the outcomes of ant–lady beetle encounters.
Mutualistic aphids are attacked by several different predators and parasitoids, which often co-occur and may compete for the same trophic resource (Schifani et al., Reference Schifani, Peri, Giannetti, Colazza and Grasso2023a). While the behavioural response of ants may also vary significantly at the individual level (Novgorodova, Reference Novgorodova2015), the differences among ant species may be crucial in favouring some arthropod natural enemies over others for pest control (Völkl et al., Reference Völkl, Liepert, Birnbach, Hübner and Dettner1996), and a better understanding of these dynamics may lead to improved biological control of several pest hemipteran species. In our experiment, the worse aphid protector, Ca. piceus, is a species with small- to medium-sized colonies that is generally described as timid, while the remaining, more effective species are characterised by either very large worker size (Ca. vagus) or very large colony size (Cr. scutellaris, D. quadripunctatus, L. emarginatus) (Seifert, Reference Seifert2018). Interestingly, in comparison to the other three, D. quadripunctatus is not traditionally described as an aggressive species. While general aggressiveness has been indicated as important, no conclusive evidence has yet emerged over natural history traits that can be used to predict the role of an ant species as a good or bad defender of its hemipteran partners, making it still important to extend the baseline knowledge to the behaviour of more species (Buckley and Gullan, Reference Buckley and Gullan1991; Novgorodova and Gavrilyuk, Reference Novgorodova and Gavrilyuk2012; Wang et al., Reference Wang, Lu, Peng and Segar2021; Schifani et al., Reference Schifani, Peri, Giannetti, Colazza and Grasso2023a).
Acknowledgements
This work has benefited from the equipment and framework of the COMP-HUB Initiative, funded by the ‘Departments of Excellence’ program of the Italian Ministry for Education, University and Research (MIUR, 2018–2022). We thank Carlo Vitali for his support in the fieldwork as a part of his internship at the University of Parma.
Competing interests
None.
