Nest usurpation: a specialised hunting strategy used to overcome dangerous spider prey

Hunting other predators is dangerous, as the tables can turn and the hunter may become the hunted. Specialized araneophagic (spider eating) predators have evolved intriguing hunting strategies that allow them to invade spiders’ webs by adopting a stealthy approach or using aggressive mimicry. Here, we present a newly discovered, specialized hunting strategy of the araneophagic spider Poecilochroa senilis (Araneae: Gnaphosidae), which forces its way into the silk retreat of the potential spider prey and immobilizes it by swathing gluey silk onto its forelegs and mouthparts. Poecilochroa senilis has been reported from the nests of a several, often large, spider species in the Negev desert (Israel), suggesting specialization on spiders as prey. Nevertheless, in laboratory experiments, we found that P. senilis has a wider trophic niche, and fed readily on several small insect species. The specialized nest-invading attack was used more frequently with large spiders, and even small juvenile P. senilis were able to attack and subdue larger spiders. Our observations show that specific hunting tactics, like nest usurpation, allow specialized predators to overcome defences of dangerous prey.

www.nature.com/scientificreports www.nature.com/scientificreports/ the secondary use of shelters. Previous research on other closely-related species 8 suggests that P. senilis might be a predator that ambushes other spiders in their retreats.
The aim of our study was to investigate the trophic ecology of Poecilochroa to determine the nature of the interaction between Poecilochroa and its hosts. We tested the hypothesis that Poecilochroa is a predator of shelter-building spiders. Alternatively, but not mutually exclusively, Poecilochroa might utilize empty shelters as retreats. First, we investigated whether its fundamental trophic niche includes spiders. We then conducted observations to reveal whether it uses aggressive mimicry or another deception strategy to usurp the nest and the host, represented by the jumping spider Mogrus logunovi Prószynski, 2000 (Fig. 1B, further shortened to Mogrus). Given the fact that the host species are larger than Poecilochroa, we anticipated the use of a specialised capture strategy.

Results
Fundamental trophic niche. Poecilochroa accepted eight out of the ten prey orders offered, but at significantly different frequencies (GEE-b, χ 2 9 = 27594, P < 0.0001, Fig. 2). Poecilochroa did not accept beetles or ants. Three prey types were accepted at a significantly lower frequency than average: woodlice, cockroaches, and crickets (Binomial tests, P < 0.04). Caterpillars were accepted at the average frequency (Binomial test, P = 0.7). Four prey types were accepted at a significantly higher frequency than average: spiders, springtails, termites and fruit flies (Binomial tests, P < 0.001). These results indicate that Poecilochroa is araneophagous, but not exclusively so.   www.nature.com/scientificreports www.nature.com/scientificreports/ Capture efficiency. The capture success of prey spiders varied significantly with the relative prey/predator size ratio (GEE-b, χ 2 1 = 34.2, P < 0.001). Poecilochroa individuals were still able to successfully capture offered wolf spiders (Pardosa sp.) in the half of the cases when a relative prey/predator (prosoma) size ratio was equal to 1.29 (i.e. wolf spiders were larger than Poecilochroa) (Fig. 3).
Nest usurpation. Poecilochroa used a repertoire of behaviours to usurp a Mogrus nest (Table 1)

. Once
Poecilochroa contacted the silk of a Mogrus nest, it either continued to move on the nest or paused. Sometimes Mogrus abandoned the nest while Poecilochroa was walking or standing on the nest. Poecilochroa occasionally plucked the nest silk, i.e. repeatedly pulling at it sharply with one or several legs, or shivered, i.e. bouncing its body with all legs in contact with the silk. Mogrus showed no visible response to these movements. Poecilochroa continued to move on the nest, accompanied by pauses, until it found one of the two nest entrances. If the nest was empty, Poecilochroa moved inside and remained there. When Poecilochroa tried to enter an occupied nest, Mogrus usually defended its nest by pulling in the silk at the entrance, and thus closing it, or by simply blocking the entrance. Poecilochroa tried to penetrate this defence by pulling or chewing the silk. If Mogrus failed to defend the nest, Poecilochroa crawled inside and Mogrus either was killed or escaped via the second entrance (Fig. 4, Video S1).
Poecilochroa individuals attempted to invade occupied Mogrus nests in 91% of all observations (N = 32). In 9%, Poecilochroa made its own silken retreat within the box away from the Mogrus nest and did not attempt to enter the nest. Poecilochroa invaded the nest and captured Mogrus in 38% of all usurpation attempts (N = 29); either immediately after entering the nest (24%), or after Mogrus escaped the nest but returned within 24 hours (14%). In 17%, Poecilochroa also invaded the nest, but Mogrus escaped and did not return within 24 hours. In 41% of usurpation attempts, Mogrus was able to defend its nest successfully, and in one case (4%) Mogrus even killed Poecilochroa (Fig. 4).

A. Behaviour of Poecilochroa Description
Contacting a nest Contacted the silk of the nest.

Locomotion
Walked on the nest or the branch.
Remaining immobile Stopped without further locomotion.
Finding an entrance Encountered one of the two nest entrances.

Pulling
Pulled the silk with legs I and/or II.
Silk chewing Chewed the silk.
Entering a nest Crawled inside the nest.

B. Response of Mogrus Description
Occupied nest? Was it present inside the nest? www.nature.com/scientificreports www.nature.com/scientificreports/ Capture success on nests declined with the relative size ratio of the prey and the predator, as larger Mogrus individuals were better able to defend the nest. In addition, it differed from the capture success with wolf spiders as prey (GEE-b, χ 2 1 = 4.2, P = 0.04), as Poecilochroa invaded occupied nests with a 50% capture success rate at a lower relative body ratio equal to 1.12 (Fig. 3). Thus Poecilochroa was less effective in capturing Mogrus than in capturing wolf spiders that were captured with a same success rate at a higher body ratio equal to 1.29.
When presented with an empty Mogrus nest, 79% of Poecilochroa individuals (N = 19) entered the nest and remained inside after the first hour, with even more individuals (95%) occupying the nest after 24 hours. predatory behaviour. Poecilochroa used a range of behaviours to subdue Mogrus after a direct contact (Table 2). When approaching Mogrus, Poecilochroa usually lunged at it or pushed it with its forelegs so that Mogrus could not lunge back at Poecilochroa. If Mogrus resisted, Poecilochroa curled its opisthosoma ventrally towards Mogrus and extruded gluey silk from its piriform glands onto the prey's forelegs and mouthparts to immobilize it (Fig. 5). Hunting sequences typically ended with Poecilochroa walking over Mogrus and biting it (Fig. 6, Video S2). The number of silk swathing events in a single hunting sequence significantly increased with the relative size of the prey (GEE-p, χ 2 1 = 11, P < 0.001, Fig. 7). Poecilochroa occasionally hunted relatively smaller Mogrus spiders without the use of piriform silk, while in several observations it applied the silk repeatedly on larger Mogrus spiders (Fig. 5). A similar silk swathing attack was also used on fleeing or resisting wolf spiders.

Discussion
We show here that the fundamental trophic niche of the spider Poecilochroa senilis is moderately wide. Spiders were one of the preferred prey types, although not the only one, suggesting that Poecilochroa is moderately stenophagous. Prey specialization among araneophages, however, is not usually strict, as they often accept several alternative prey types beside spiders, though at lower frequencies [16][17][18][19][20] . We investigated the trophic niche in juveniles of P. senilis only, because the number of adult individuals collected in the field was very low for the experiment. Although the trophic niche breadth may increase with age of spiders as bigger spiders can hunt bigger prey 21 , specialized spiders hunting dangerous prey like ants or other spiders are usually able to subdue larger prey than themselves even as juveniles [22][23][24] . Similarly, Poecilochroa was able to overcome larger spiders than itself as a juvenile. Its hunting strategy towards spider prey was therefore very effective, indicating prey-specialised hunting behaviour.
In nature, in addition to the nest-building salticid M. logunovi, Poecilochroa was found to attack and feed on two species of web-building spiders by capturing them in their nests, namely the widow spider Latrodectus  (Table 1) and a transition matrix based on the 'Nest usurpation' experiment.
www.nature.com/scientificreports www.nature.com/scientificreports/ revivensis and the cribellate spider Stegodyphus lineatus 15 . These observations support our laboratory experiments indicating that Poecilochroa is able to handle a rather broad range of large and dangerous spider prey. The extent to which it feeds on insects in nature is unknown. However, spiders are especially numerous arthropods in desert ecosystems, both in terms of abundance and biomass [25][26][27] . Intraguild predation (predation among predators) plays an important role in shaping the composition of desert arthropod communities 27,28 , as more than 50% of the diet of desert predatory arthropods is composed of other predators; and predatory arthropods constitute a high proportion of all desert arthropods 29 . We collected Poecilochroa solely in association with other spiders, suggesting that it is locally specialized on these abundant prey in the desert.
We found that Poecilochroa is able to subdue Mogrus inside the nest. Entering a spider web or nest is dangerous, as it is an extension of the spider's senses and any intruder can become a prey. Web-invaders usually use several approaches to avoid this: they either deceive their victim by means of aggressive mimicry [8][9][10][11] , by approaching the prey stealthily 12,13 , or by leaping suddenly onto webs from a distance 30,31 . Poecilochroa used a very different approach -namely, rapid nest usurpation: once it found the entrance of the nest, it tried to enter and swiftly subdue the resident spider without using stealth. A similar hunting tactic was observed in the spider Nyssus coloripes Walckenaer, 1805 (Corinnidae), which relied on its rapid, erratic style of locomotion when invading webs 32 , and Zelanda erebus (L. Koch, 1873) (Gnaphosidae), which was able to chew through silk and force its way inside the   www.nature.com/scientificreports www.nature.com/scientificreports/ nests of jumping spiders. However, Z. erebus also utilized aggressive mimicry when invading webs 8

. Although
Poecilochroa sometimes moved its body and legs during locomotion or when standing on the nest in a way that may produce vibrations, Mogrus was not lured out of the nest in search of a prospective prey. However, as Mogrus sometimes left the nest before Poecilochroa invaded it -that is, after it had become aware of Poecilochroa's presence (Michálek, pers. obs.) -such leg movements may have been a way of checking whether Mogrus was present in the nest. We cannot, however, rule out the possibility that Poecilochroa utilizes other tactics when invading the webs of other species. In this respect, we tried to observe invasions of the nests of adult Stegodyphus lineatus, but Poecilochroa did not attack these spiders in the web at all (Michálek, pers. obs.), possibly because the S. lineatus individuals were too large. Nevertheless, Poecilochroa may shift to attacking S. lineatus during the season when these spiders are juveniles 15 .
Poecilochroa was not always successful in invading nests. It was able to subdue large spiders when hunting outside nests, but was less successful in capturing them in their shelters. As a key element in overcoming large prey is immobilization with piriform silk, it seems that Poecilochroa is limited in the use of this silk when entering the narrow Mogrus nest. A conditional strategy for hunting larger Mogrus spiders may be to wait outside the nest and ambush Mogrus when it leaves the nest. However, the prey may more easily escape in the open space outside the nest. Alternatively, Poecilochroa may enter an empty Mogrus nest while Mogrus is foraging and ambush Mogrus when it returns, as we also observed.  www.nature.com/scientificreports www.nature.com/scientificreports/ When a resident Mogrus escaped and was not captured later, Poecilochroa often remained inside the host nest at least for one day. It is possible Poecilochroa utilize empty shelters even for longer periods, as we found Poecilochroa exclusively inside Mogrus nests during our field survey. Usurping webs or nests may have other benefits 8 ; in this case, Poecilochroa may utilize the Mogrus nest as a shelter, feed on the eggs or young of the resident spider (Michálek, pers. obs.), or use the nest for its own oviposition 15 .
Poecilochroa utilized gluey piriform silk to immobilize spider prey. Web-building spiders use piriform silk as attachment discs 33 , but gnaphosid spiders possess a modified spinning apparatus allowing them to use piriform silk as adhesive tape and to hunt dangerous prey, such as other spiders 34 . We observed that Poecilochroa used silk conditionally depending on the relative prey size. Poecilochroa was able to subdue relatively smaller spiders without the use of swathing silk, while it applied silk repeatedly when immobilizing larger spiders. It thus showed a high level of versatility on prey type and size. Specialized spiders often display stereotyped hunting behaviour 23,24 . Several araneophagous spiders, however, utilize a variety of tactics depending on the predatory context 8,32,35 . As spider-eating spiders are usually not as strictly specialized as ant or termite specialists 36,37 , araneophagous spiders may retain greater plasticity in their hunting behaviour. Some specialized spiders have an innate search image of their focal prey, or the search image is formed in a single encounter 38,39 . Poecilochroa readily attacked and preyed upon Mogrus even as a naïve juvenile, when this prey type was novel to it, and it was already able to subdue large prey.
A flexible line of defences from multiple primary to secondary strategies has evolved in prey that allow them to avoid or deter predators 40 . Anti-predator strategies may be also influenced by the individual's condition, for example, its reproductive state 41 . Therefore, a predator is confronted with a diversity of prey defensive mechanisms. A successful predator must have a versatile predatory behaviour that allows it to overcome all subsequent defences of a prey. This is even more pronounced when the prey is a potential predator as well. Here, we have shown Poecilochroa maintains such versatility by overcoming both primary and secondary defences of its dangerous spider prey: it penetrates the defended shelter by pulling and chewing silk, ambushes the spider fleeing from shelter, usurps and utilizes the empty shelter, and immobilizes dangerous, resisting prey with silk. Methods studied species. During our survey in the Negev desert, we found nests of the jumping spider M. logunovi occupied by Poecilochroa (Michálek & Pekár, pers. obs.). Therefore, we focused on the interaction between these two species. Poecilochroa individuals (Fig. 1A) were collected at Mashabim (31°00′07.3″N, 34°45′18.3″E) and Retamim (31°06′27.5″N, 34°39′15.0″E) sand dunes in the Negev desert, Israel, in April 2016 and in March and April 2017. Individuals were collected by inspecting nests of Mogrus found on different shrub species (mainly Retama retama, Artemisia monosperma, and Thymelaea hirsuta). Mogrus individuals (Fig. 1B) were collected in 2017, along with their nests, on shrubs at the same localities as Poecilochroa. Mogrus nests are composed of several layers of silk, forming a 'sleeping bag' with two openings, one at each end (Fig. 1B). Only 1.4% of the 510 Mogrus nests examined during March and April 2017 were found to contain Poecilochroa spiders. Some females of Poecilochroa collected in the field produced egg-sacs in the lab. Hatched juveniles were also used in experiments along with individuals collected in the field.
Poecilochroa spiders were kept in plastic vials (length 55 mm, diameter 12 mm) containing moistened gypsum, and stored in a chamber at constant temperature (22 ± 1 °C) and under a 16:8 LD regime. Mogrus spiders were kept in plastic containers (55 × 55 × 75 mm) along with their nests at room temperature (22 °C) and under a natural 14:10 LD regime. All spiders were fed at least once a week with Drosophila flies ad libitum or were allowed to consume the prey accepted in laboratory trials to standardize their satiation level. Prey used in experiments were either laboratory reared or collected around the university campus in Brno, Czech Republic and kept at 10 °C before using in the experiment. Experiments were performed from August 2016 to April 2018 in Israel and in the Czech Republic. All statistical analyses were performed in the R environment 42 .

Fundamental trophic niche.
To investigate the fundamental trophic niche of Poecilochroa, prey acceptance experiments 43 were performed with 39 juvenile Poecilochroa individuals. Spiders were starved for one week before being used in trials. Individuals were placed singly in Petri dishes (diameter 50 mm). The trials began after at least 1 h of acclimation. Ten prey types (Table 3) were offered to each spider in a randomised order. Each prey type was offered to each individual spider only once. If the prey was not attacked within one hour, it was replaced with a different prey type. The trial ended when a spider had killed and consumed a prey. If a spider did not accept any prey type, it was considered to be unmotivated to eat (e.g. satiated or preparing to moult) and data from such trials were rejected. Trials were performed at one-week intervals for each individual.
The differences in acceptance rates for ten prey types were analysed using Generalised Estimating Equations (GEE) with binomial errors from the geepack package 44 . GEE is an extension of the Generalised linear model (GLM) for correlated data. It was used because there were repeated measurements on each individual spider 45 . An autoregressive correlation structure (AR1) for replicated observations over time was used to account for these temporal replications. Subsequently, the relative frequency of the acceptance of each prey type was compared to the average prey acceptance for all prey types using a binomial test. The standardized Levins' index (B A ) of niche breadth 46 was used to calculate the fundamental trophic niche breadth. Values of B A higher than 0.6 indicate a wide niche; values below 0.4 indicate a narrow niche 47 . efficiency of prey capture. To measure the capture efficiency for differently sized prey, wolf spiders (Pardosa sp.) of various sizes and developmental stages were randomly offered to Poecilochroa in a similar manner to the previous acceptance trials. Individuals of Poecilochroa were placed singly in Petri dishes and offered the prey item after 1 h of acclimation. If the prey was not accepted within one hour, it was replaced by a smaller one (on average two times smaller than the previous prey item). If a spider did not accept smaller prey, it was considered to be unmotivated to eat and such data were discarded. The length of the prosoma in all spiders was www.nature.com/scientificreports www.nature.com/scientificreports/ measured under a LEICA EZ5 binocular lens with an ocular micrometer. In total, 84 trials with 37 juvenile individuals of Poecilochroa were performed. The logit model with binomial distribution using GEE was used to fit the binary data, as there were repeated measurements on each individual spider. An AR1 correlation structure was used to account for the temporal replications.
In addition, 45 trials using freshly hatched first instar juveniles of both Poecilochroa and Mogrus were performed. One week before the experiment, each spider was fed with a Drosophila fly. In each trial, two individuals, one of each species, were placed in a Petri dish (diameter 35 mm) and observed until one of the spiders was killed by the other. Subsequently, the hunting success of Poecilochroa or Mogrus was recorded. The size of the prosoma of all juvenile spiders was measured under a LEICA EZ5 binocular lens with an ocular micrometer.

Nest usurpation. To find how Poecilochroa penetrates jumping spider nests, interactions between
Poecilochroa and Mogrus were staged on Mogrus nests. Female Mogrus and Poecilochroa spiders were fed five days before the trials with Drosophila flies ad libitum. Occupied nests of Mogrus on their supporting branches were placed in transparent plastic boxes (55 × 55 × 75 mm) and acclimated for at least 24 hours. Then, a female or juvenile Poecilochroa was introduced on the branch above the nest and the interactions between the two spiders were recorded on a camcorder (Canon Leigra HF R56). The recording ended one hour after the introduction of Poecilochroa. If Mogrus was not captured within 1 hour, the outcome of the interaction was recorded the following day, after an additional 23 hours. In total, 32 observations were made on occupied Mogrus nests and the behaviours of Poecilochroa and responses of Mogrus were described ( Table 1).
The effect of the relative sizes of the prey and predator on usurpation success was analysed using GEE. The logit model with binomial distribution and an AR1 correlation structure was used to account for temporal replications, as several Poecilochroa individuals were used more than once. The capture efficiency on Mogrus was compared with that of wolf spiders as prey.
In addition, 19 observations were conducted of Poecilochroa behaviour in response to empty Mogrus nests. The nests were placed in transparent plastic boxes (55 × 55 × 75 mm) and the resident Mogrus spiders were removed. Then, female or juvenile Poecilochroa spiders, fed five days before the trials with a surplus of Drosophila flies, were introduced into the boxes with empty nests. The presence of Poecilochroa inside or outside the nest was documented after one hour and 24 hours. predatory behaviour. To observe how Poecilochroa can overcome spider prey, predatory encounters between Poecilochroa and Mogrus were staged. The mean prosoma size ratio of Mogrus and Poecilochroa spiders in this experiment was 1.00 ± 0.14. Spiders were fed five days before trials with Drosophila flies ad libitum. At the start of each trial, spiders were put individually into small elongated glass tubes (length 35 mm, diameter 6 mm) to simulate the narrow space inside a Mogrus' nest. Then, the two tubes, one hosting Poecilochroa and the other hosting Mogrus, were connected by their openings and the interaction between the spiders was video-recorded. If the interaction did not result in predatory behaviour within 30 minutes, the trial was ended. Hunting sequences were recorded either on a camcorder (Canon Leigra HF R56, 22 trials) or BW high speed camera (IDT MotionXtra N3, 15 trials) at 50 FPS, to record the details of the attack. In total, 34 complete hunting sequences out of 37 recordings were obtained. In the complete hunting sequences, the different behaviours of Poecilochroa were distinguished ( Table 2).
The effect of the relative sizes of the prey and predator on the number of "silk swathing" events in individual sequences was analysed using GEE with Poisson distribution and an AR1 correlation structure to account for temporal replications, as several Poecilochroa individuals were used more than once. Piriform silk on captured Mogrus individuals was visualized by means of fluorescent dye following the protocol by Johnson et al. 48 .  Table 3. List of prey types used in acceptance experiments, their body sizes (*body size, **prosoma size), and the number of Poecilochroa individuals to which was offered given prey (N).