Introduction

There are many benefits for individuals living in groups, including the increase in access to resources due to cooperative foraging (Clark and Mangel 1986; Packer and Ruttan 1988; Ward and Webster 2016), defence against predators, increase in mating opportunities (Majolo et al. 2008), and alloparental care (Guindre-Parker and Rubenstein 2018). Behavioural interactions between individuals within these groups vary in complexity, ranging from interattraction and coordinated movements to avoid predation (Krause and Ruxton 2002) to cooperative hunting involving division of labour with role specialization (Gazda et al. 2005). Within this range, social interactions were previously reported in several taxonomical groups, such as insects, spiders, fishes, birds, and mammals (Brown and Brown 1996; Majolo et al. 2008; Ward and Webster 2016). Nonetheless, group living may also have costs, such as increased interindividual competition, which may become so strong that losses outweigh the advantages of group foraging (Majolo et al. 2008). Thus, group size and intragroup interactions involved in mutual tolerance usually depend on several factors, such as food availability, size of available prey, predation and parasitism pressures, and even specific abiotic conditions (Guevara et al. 2011, Zöttl et al. 2013, Creel et al. 2014, Hoffman and Avilés 2017).

In spider social structures categorized as ‘colonial’, individuals form permanent or long-term aggregations with each spider defending its territory. The interconnection of several individual orb-webs in these colonies may increase individual foraging success due to the ricochet effect (Lubin 1974; Uetz 1989). Furthermore, organization in colonies can facilitate the use of habitats which are not available to solitary individuals (Lubin 1974) and may decrease the per capita amount of silk required for web construction (Lloyd and Elgar 1997). Another potential benefit of colonial organization is the decrease of predation risks (Rayor and Uetz 1990; Uetz and Hieber 1994). This is due to predator warning signs which can be transmitted over interconnected webs (Uetz et al. 2002) and due to the dilution effect in large groups (Uetz and Hieber 1994). Finally, small and young individuals may be greatly favoured by cooperative hunting. Collective prey capture by those individuals may increase prey availability, allowing the immobilization and consumption of large and/or aggressive insects. This advantage may lead to the suppression of agonistic behaviours between group members at least during part of their lives.

The long-lasting maintenance of the collective web structure in colonial spider species, however, may be constrained or compromised by frequent agonistic interactions between individuals. Agonistic interactions within colonial groups usually occur due the lack of space for individual webs or due to competition for particular sites for web construction (Rayor and Uetz 2000; Alves-Costa and Gonzaga 2001; Wenseleers et al. 2013). Furthermore, agonistic interactions between two or multiple individuals may occur during prey capture events (Lubin 1974; Yip et al. 2017), with the outcome usually depending on the body size of the competitors (e.g. Hodge and Uetz 1995). Indeed, spider size (or instars) may have a fundamental role in determining interactions between neighbours in these colonies. Large spiders may be able to catch and subdue large prey without cooperation, actively defending their territories (orb-webs) against the invasion of other group members. Thus, possible advantages of prey monopolization and frequent agonistic interactions when spiders reach the late instars may contribute to colony dissociation, compromising the nutritional benefits of remaining associated with nestmates.

Another important trigger for interactions in colonies is probably the escaping behaviour of insects after being intercepted by web threads. Distinct vibratory patterns produced by insects may result in a distinct number of spiders attracted, probably influencing the frequency of both kinds of interactions, conflicts and cooperation. Flies and moths intercepted in colonial webs of Anelosimus eximius (Keyserling, 1884) (Theridiidae), for example, usually vibrate their wings for a long time after interception, and these movements attract many spiders (Souza et al. 2007). Moths, in addition, often move away from the interception site, promoting intense vibrational stimuli. The displacement occurs because the scales of their wings flake off after the initial contact with the web, covering the glue droplets of adhesive threads (Nentwig 1982, Diaz et al. 2018). Grasshopper nymphs, on the other hand, usually move only their bodies and legs trying to escape. This behaviour produces erratic and non-rhythmical movements with low potential to propagate to nearby webs (Endo 1989, Souza et al. 2007).

The spider Parawixia bistriata (Rengger, 1836) (Araneidae) is a good experimental model for the study of the co-occurrence of cooperation and competition in colonies. It can be categorized as a periodic-social territorial species (Avilés 1997). Individuals remain in groups for most of their lives and disperse as subadults and/or adults. In this study, we describe the occurrence of these interactions during prey capture of young and relatively older individuals. We hypothesise that cooperation will be more frequent in November, among young juveniles, and conflicts will be more frequent in December. We also experimentally tested the effects of two different prey types (Lepidoptera and Orthoptera) of different sizes on the interactions between spiders. Our hypotheses are as follows: (1) Moths will attract more spiders to the interception site and elicit cooperative behaviours, once prey monopolization is almost impossible in situations involving several individuals; (2) grasshoppers will be captured by the resident or attract only the closest neighbour, eliciting more conflicts; (3) heavier prey will attract more spiders during prey capture events; and (4) large spiders will win most conflicts, regardless of residence.

Material and methods

Study area

This study was conducted in the Panga Ecological Station (19°11′40″ S, 48°19′06″ W), a 409 ha reserve covered predominantly by Brazilian savannah (Cerrado sensu stricto), located 30 km south of Uberlândia, Minas Gerais State, Brazil. The region is characterized by a tropical climate with two distinct seasons, a dry winter (from May to September) and a rainy summer (from October to April). Average annual rainfall varies from 1400 to 1700 mm, and monthly temperature varies from 8.5 to 33 °C (average 22.8 °C) (Paiva et al. 2007).

Study species

Parawixia bistriata is a colonial orb-weaving species, which can be found in savannahs and open vegetation biomes in Brazil, Bolivia, Paraguay, and Argentina (Levi 1992; World Spider Catalog 2019). The species has an annual life cycle, and juveniles mature from September to March in their natal groups (Sandoval 1987). Juveniles usually construct individual orbicular webs after sunset (Fowler and Gobbi 1988a), and these webs are attached to supporting threads constructed and maintained by several members of a group (Sandoval 1987). Individual webs are defended as exclusive feeding territories. However, as the individual webs are interconnected and vibratory signals of intercepted prey are transmitted between webs, this collective structure favours the occurrence of interactions during prey capture events, especially when the prey intercepted is relatively large (Fowler and Gobbi 1988b; Campón 2007).

Observation of prey interception and capture

We observed segments of the colonies containing more than three spiders with interconnected individual webs, always from 19:00 to 01:00 h, in November and December (rainy season) of 2016 and 2017. We conducted a total of 17 h of observations (8.5 h each year), using 9 colonies (2 in the first year and 7 in the second). Observations were divided in 30-min samples, each one conducted in a different subgroup of spiders. During this time, we recorded prey composition (insect orders) and body size (total length, in mm) of all insects captured by the spiders. We excluded from records all the insects intercepted by the webs but ignored or removed by the spiders. For each capture, we recorded the number of spiders interacting and their behaviour, including aggressive movements (bites, abrupt leg movements, attempts to pull parts or the entire prey away from other individuals) and cooperation (spiders simultaneously biting and/or consuming the prey at the interception site). Lanterns used during all the observations were covered in red cellophane paper to minimize insect attraction. We used Fisher’s exact tests to evaluate whether the frequencies of cooperation and conflict during prey capture vary between months.

Experimental feeding

We carried out a prey offer experiment in January and February 2017 to estimate the frequency of cooperation and agonistic interactions among individuals with neighbouring webs. We conducted the experiment using colonies with subadult individuals of P. bistriata and two orders of insects (Lepidoptera and Orthoptera) as prey items. This experiment was designed to also evaluate the effect of prey mass in the interactions. We collected the insects in the field using active search and a light trap, and these insects were kept in vials and offered to the spiders within 2 days of collection. All insects used were in good conditions when offered to the spiders. We weighted all insects immediately before the experiment using an analytical balance (Shimadzu Model AU, at a precision of 10 μg).

For prey offer, we selected areas of the colonies with at least three individuals with interconnected webs. We marked the individual located in the central web and the two nearest neighbours, using different enamel paint colours, by softly touching their abdomens with a small brush while they were in the centre of their webs. Spiders usually remain motionless during manipulation, only assuming a defensive position by contracting their legs for a few seconds and returning to the normal waiting position shortly afterwards. We placed the insects in the intermediate position between the hub and the edge of the orb. After placing the insect in the web, we recorded the number of spiders that moved to the interception site.

During prey capture, two or more neighbours can cooperate or engage in agonistic interactions to monopolize the insect. In the interactions categorized as cooperative, the spiders attacked prey simultaneously, without showing aggressive behaviours or attempts to move the prey away. We considered interactions between them as conflicts when a direct contact between the spiders involved was observed, leading to prey monopolization by one contestant. These contacts involve mutual touches of legs and/or chelicerae and also attempts to move the prey away from the contestant, usually resulting in prey monopolization by one of the spiders (considered as winner). Losers flee from the web without resources. When conflicts occurred, we collected and weighted all involved individuals after the interactions. After this procedure, we reinserted the spiders in their original position within the colony.

To test whether cooperation and conflicts during prey capture events depended on prey mass and type, two logistic regression mixed-effect models were fitted, exploring the effect of each variable and their possible interactions. Colonies were considered as a random factor in these analyses. In the first model, we considered the probability of cooperation (1 = cooperative capture, 0 = solitary capture or conflict) and in the second the probability of agonistic interactions (1 = direct conflict, 0 = solitary or cooperative capture). When we observed cooperation, the number of spiders capturing (those that moved to the capture site and actively participated in the immobilization process) and the number of spiders feeding on the immobilized prey were used in a linear regression model as a function of prey mass. All statistical analyses were performed using R software, version 3.3.3 (R Development Core Team 2016).

Results

Observation of prey interception and capture

During observations, we recorded the interception of 181 insects (Fig. 1a) of nine orders: Diptera, Coleoptera, Hemiptera, Lepidoptera, Blattodea, Hymenoptera, Orthoptera, Neuroptera, and Mantodea. Most insects captured were relatively small Diptera and Coleoptera (Fig. 1b). These small items were usually captured by the orb-web resident, without interactions with other spiders. In November, we observed more events of cooperative captures (P = 0.002, Table 1), while the frequency of conflicts did not vary between months (P = 0.587, Table 1) Fig. 2.

Fig. 1
figure 1

Insects captured by P. bistriata (Di, Diptera; Co, Coleoptera; He, Hemiptera; Le, Lepidoptera; Bl, Blattodea; Hy, Hymenoptera; Ne, Neuroptera; Or, Orthoptera; and Ma, Mantodea). a Proportion of each prey type captured. b Body length of prey captured

Full size image
Fig. 2
figure 2

Capture events. a Young individuals prey catching cooperatively (in November). b Cooperative capture by older individuals (in January). c Conflict and dispute for prey. d Subadult individual monopolizing prey with body size superior to her own

Full size image
Table 1 Pooled number of prey items captured in 2016 and 2017 during behavioural observations
Full size table

Experimental feeding

Insects used in this experiment exhibited distinct behaviours when captured in webs. Moths typically flapped their wings until they were immobilized (see video 1, cooperation), which is why their presence was rapidly detected by neighbouring spiders. Grasshoppers remained motionless or only moved their legs (see video 2, agonistic interaction). Low-mass prey, especially grasshoppers, tangled in the web threads occasionally ceased to move, whereas medium and large prey only stopped moving after the action of one or more spiders.

We offered a total of 78 insects, 39 moths (Mass(g): \( \overline{\mathrm{X}} \) ± SD = 0.376 ± 0.363) and 39 grasshoppers (Mass(g): \( \overline{\mathrm{X}} \) ± SD = 0.345 ± 0.320). Of the 78 capture events, 12 (15%) occurred as a result of cooperative capture. However, 11 of these cooperative captures occurred with moths and only one with grasshoppers, leading to the exclusion of this last group from the analysis in cooperative interactions. The probability of cooperation during capture was affected by moth mass (χ2 = 9.827, df = 1, P = 0.001, Fig. 3). The number of spiders involved in prey capture did not correlate with the mass of moths (F1,9 = 2.278, P = 0.165, R2 = 0.202, b1 = 3.117, SE = 2.065; Fig. 4a). However, the number of spiders taking part in collective feeding after capture was influenced by this variable (F1,9 = 14.046, P = 0.004, R2 = 0.609, b1 = 9.449, SE = 2.521; Fig. 4b).

Fig. 3
figure 3

Probability of occurrence of collective captures of moths according to prey mass

Full size image
Fig. 4
figure 4

Correlations between prey mass. a The number of spiders taking part in prey capture. b The number of spiders consuming prey after immobilization

Full size image

(MP4 22386kb)

(MP4 22103 kb)

Of the 78 capture events, 27 (34%) produced conflicts between spiders in adjacent webs (18 using grasshoppers and 9 using moths). In moths, the frequencies of conflicts did not correlate with prey mass (Table 2, Fig. 5a); in grasshoppers, however, conflict frequencies were correlated with this variable (Table 2, Fig. 5b). Only 11 conflicts (41%) were won by the residents. The mass asymmetry between the contestants affected the probability of victory (χ2 = 5.206, df = 1, P = 0.022, Fig. 6), with heavier spiders winning 89% of the conflicts.

Table 2 Results of the mixed-effects logistic regression model on the effects of mass and order of the prey on the occurrence of conflicts during prey capture
Full size table
Fig. 5
figure 5

Probability of occurrence of conflicts according to prey mass. a Moths. b Grasshoppers

Full size image
Fig. 6
figure 6

Probability of victory in contests considering the difference between mass of the resident and invader

Full size image

Discussion

Parawixia bistriata has a relatively unusual type of social structure for spiders, with juveniles cooperating in foraging activities (construction of a communal framework to support the orb webs, prey immobilization, and prey consumption) but maintaining individual territories. Cooperative capture events are dependent on both, prey type and prey mass, being more frequent with moths (and possibly with other insects with similar behaviours when intercepted, such as termites), as they tend to attract more spiders. In such situations, prey monopolization is not possible, and several spiders usually take part in consumption. These cooperative events are especially common among young juveniles. Most insects intercepted by the webs, however, are individually captured and consumed by the residents, and in some cases, the attraction of just one individual, usually from the adjacent web, may result in a conflict over the prey item. Contests between the owners of adjacent webs for intercepted insects are generally won by the heavier spider, and their occurrence also depends on prey type and, at least for grasshoppers, prey mass.

Solitary orb-weaver spiders present a large diversity of foraging strategies, from stenophagy (especially regarding the capture of Diptera and Lepidoptera) (Pekár et al. 2011) to euryphagy (e.g. Murakami 1983; Wise and Barata 1983; Ibarra-Nuñez et al. 2001; Ceballos et al. 2005). There is relatively less information on diet breadth of colonial orb-weavers, but the interconnection of several orbs and the possibility of cooperative interactions among colony members may improve webs efficiency in intercepting and retaining a wide range of prey types and sizes (Uetz 1989) and also the ability of spiders to immobilize and consume large insects (Binford and Rypstra 1992, Masumoto 1998). In the present study, we observed that P. bistriata captured nine orders of insects, with a large variation in prey body size. At least part of this range is probably determined by the occurrence of frequent cooperative interactions, which allows the immobilization of relatively large insects even by young juveniles. Moreover, the wide diet range of P. bistriata is probably influenced by habitat occupation, with individual webs positioned from close to the ground to several meters above. It is possible that positional variability of webs within the colony contributes to the interception of various insects which mostly occur at specific heights (see McReynolds and Polis 1987; McReynolds 2000).

In P. bistriata, cooperation in prey capture seems to depend on the behaviour of intercepted insects. The probability of cooperative capture and the number of spiders attracted during the phases of prey immobilization and feeding were influenced by prey type. Despite the interconnected webs, maintenance of individual orbicular structures is likely to restrict the propagation of vibrations over long distances. Thus, only certain patterns (or intensities) of vibration may attract individuals located in adjacent territories. This may explain the rare occurrence of cooperation in capturing Orthoptera. Although these insects produce strong vibrations with sudden movements of the hind legs, these movements are characterized by lower frequencies, compared with the wing flapping of Diptera or Lepidoptera (Souza et al. 2007). The vibrations of large or medium moths, however, are probably strong enough to be perceived by several spiders on adjacent webs. After immobilization, the movements of spiders feeding on the prey attract other individuals, and their number at this phase depends on prey mass.

The frequency of cooperative prey capture is higher in young juveniles. Three factors may explain this observation: (i) the maturation stage of the spiders, (ii) their size and (iii) the occurrence of flying termites and other relatively large-winged insects in the beginning of the rainy season. First, there is a tendency for greater tolerance for individuals at early developmental stages (e.g. Trabalon et al. 1996). According to Trabalon (2013), in some species, lipid alterations in the cuticle composition occur during the development, followed by an increase in the frequency of agonistic interactions between members of the group (Trabalon 2013). It is possible that this also occurs in P. bistriata, and this characteristic may be important to determine the dispersion of adults. Regarding the size of spiders, individuals in early developmental stages can obtain a greater benefit by participating in collective capture events. Cooperative behaviours may significantly increase the range of available prey and allow immobilization of relatively large insects (Campón 2007). With spider growth, individuals become more efficient in solitary captures, including the immobilization of prey which is larger than the spider, potentially decreasing the value of collective capture. Finally, termites and other large insects (considering the size of young juveniles) are a predictable and abundant resource in the beginning of the rainy season in Brazilian Cerrado (Sandoval 1987). Exploitation of these resources by small spiders probably requires the joint action of several individuals.

In most territorial colonial spiders, prey capture is essentially a solitary activity, and the occurrence of conflicts over intercepted insects seems to be rare (e.g. Lubin 1974). However, the easy access to adjacent webs and possibility of web invasion in P. bistriata seem to promote the occurrence of conflicts over prey monopolization. The higher frequencies of conflicts over Orthoptera than Lepidoptera indicate that these interactions to monopolize prey depend on the number of spiders that are attracted to the capture site. Success in monopolization is only possible when one individual is in competition with one or few others. When a moth attracts a substantial number of neighbouring spiders, it is difficult for any individual, whether resident or invader, to keep others at bay. These observations are in accordance with the ‘resource defense theory’ (see Brown 1964, Grant 1993), which predicts that the likelihood that aggressive interactions will occur is correlated to a resource’s economic defendability. Conflicts involving aggressive behaviours, however, may also be influenced by other variables, such as the occurrence of recurring interactions between the same individuals (Dubois and Giraldeau 2003) and the degree of spatial clumping of resources (Dubois et al. 2003). Those variables were not considered in our experiment, and further studies specifically focused on conflicts are necessary to investigate their effects on the frequency and the outcomes of disputes over prey items in colonies of P. bistriata and other colonial spiders.

In conflicts during prey capture, larger individuals of P. bistriata mostly had an advantage, regardless of whether the prey had been intercepted in their own or in an adjacent web. Conflicts during or after prey capture were previously reported in non-territorial social species; there, the agonistic interactions typically occurred when individuals attempted to move the whole prey item or parts of it to their refuges (e.g. Ward and Enders 1985). Although conflicts are not very common in territorial species (Wise 1983), agonistic interactions can occur more often when prey interception by a particular web is very frequent (Gan et al. 2015). This is in accordance with the results of several studies demonstrating that aggression increases as resources became aggregated in space (Golgberg et al. 2001, Dubois et al. 2003). Regarding the defence of territories, the conflict in spiders is typically won by the larger individual or by the resident, when the competitors are of similar size (Hodge and Uetz 1995, Schuck-Paim 1999). However, the residence status did not seem to have an effect on the outcome of conflicts during prey capture in this study, as even when the competitors were of similar size, residents frequently lost the prey. The reasons for the advantage of the heaviest spiders in contests must be evaluated in a further study focusing specifically on the behaviour routines during agonistic interactions and on the costs for the resident in being removed from its original web.

A final aspect to be considered is the spatial distribution of orbicular webs over the supporting threads (the number of adjacent interconnected webs). This variable may influence the frequency of cooperative and agonistic interactions because close proximity of webs implies a more efficient propagation of vibrational signals and, consequently, attracts individuals to sites of prey interception. This may also explain the higher frequencies of cooperation when individuals were smaller and constructed relatively small webs in close proximity.

Concluding, our study emphasized the importance of prey types and sizes on the frequency of intracolonial interactions in P. bistriata (and possibly other colonial species) and showed that cooperation during foraging activities may be more important to young juveniles, decreasing its frequency in later instars. The advantages of cooperative hunting during the initial instars may be important even for some solitary species (Pekár et al. 2005, Bertani et al. 2008), influencing the range of insect sizes included in their diet and, consequently, food availability. The maintenance of tolerance and cooperative behaviours in later instars probably allowed the establishment of complex societies in spiders. In the case of P. bistriata, however, these advantages are progressively suppressed by the ability of large individuals to capture profitable prey by themselves and to steal prey from their neighbours. Finally, our results indicate that it is also important to consider the behaviour of trapped insects after interception as a promoting factor for interactions.