Individuals of the social spider mite Stigmaeopsis longus live communally in narrow, humid nests made from silk threads and display nest sanitation behaviour through the coordinated deposition of faeces. We used artificial dust to experimentally determine that females of this species use silk threads to perform regular cleaning of the nest space and eggs. We first learned that silk-weaving behaviour is not a by-product of nest building (nest reinforcement); rather, it is actively performed as a function of cleaning the living space and eggs. Furthermore, we determined the effectiveness of the attending females by artificially manipulating their natural habitat, which is clearly connected to the cleaning behaviour by parental females. As such, we have uncovered an extraordinary new role of silk threads as devices for cleaning the nest space and/or eggs. These results strongly indicate that special adaptations for maintaining clean habitats are essential for animals to evolve aggregative social lives.
Group living in animals confers several advantages for improving individual lives, such as social organization and anti-predation effects (i.e. extended parental care and group defence [1–4]). Group living also involves inherent disadvantages, such as risks from mass disease infection, waste production and rapid resource deterioration [5,6]. In relation to these, the importance of faecal-load-reducing mechanisms (larval blind gut) in the evolution of eusociality has been recently discussed with regard to Hymenoptera . A near-total absence of larval faeces in the nests of hymenoptera could facilitate sociality because of the reduced workload necessarily allotted to waste management. Human beings living gregariously in urban areas face a similar array of problems .
Among social insects such as earwigs, ants and termites, there are several well-investigated examples where parental (or worker) care behaviour can successfully protect eggs against infection by fungi or micro-organisms [9,10]. Female earwigs periodically secrete antibiotics from their mouthparts to sanitize egg surfaces. Without such care, eggs would seldom survive in humid habitats . Worker termites living in self-constructed nest chambers secrete saliva with antibiotic properties onto eggs, thus enhancing egg survival [11,12]. As such, animals that inhabit humid and fungi/bacteria-abundant habitats and have more or less social and gregarious lifestyles can be expected to have evolved specialized care behaviours for their offspring, especially at the egg stage, to improve their survival. In other words, these reports strongly suggest that social and/or aggregative lifestyles are a prerequisite for developing behaviour for maintaining clean living spaces.
The social spider mite species Stigmaeopsis longus (Saito) (Acari: Tetranychidae) uses dense silk threads to build nests and live gregariously over depressions in the under-surfaces of host leaves (figure 1). Adults show strong counter-attack behaviours against predators and often succeed in defending their nests and offspring by driving away or even killing nest intruders . Furthermore, they display waste-management behaviours: they establish a ‘toilet’ at their nest entrance, and all nest members deposit faeces there following rules realized through tactile and chemical cues . The waste-management behaviours of S. longus undoubtedly function to prevent nest space and food (leaf surfaces) from becoming fouled through group living. However, since the inside of the closed capsule-like nest is highly humid, some additional forms of nest-space cleaning behaviours are expected from S. longus.
We first determined that S. longus females ordinarily use silk threads to perform cleaning of the nest space and eggs. These nest-cleaning behaviours are not a by-product of nest building (nest reinforcement); rather, they are performed as a function of cleaning the living space/eggs. After confirmation of the above, we determined the effectiveness of the attending females in performing this behaviour by artificially manipulating their natural habitat.
2. Material and methods
The cooperative social spider mite species S. longus naturally occurs on Sasa senanensis Sieb., in small woods on the campus of Hokkaido University, Sapporo, Hokkaido, Japan. The mites used in our experiments were collected on 15 August 2007. They were reared on the under-surface of detached S. senanensis leaves (collected from the same woods) with the under-surface turned upward in a climate-controlled chamber at 25 ± 2°C and at 50 to 70 per cent r.h. room humidity with a photoperiod of 15 L : 9 D. Laboratory experiments were carried out under the same temperature and humidity conditions, except for the photoperiod, as described later.
A detached leaf under-surface facing upward was used for the present experiments except for experiments in the field because there were several difficulties related to observing and manipulating leaf under-surfaces when facing downward. Although it has been empirically determined by mite researchers that such a reversed position of the leaf has little effect on mite life history and behavioural performance , it is necessary to check the effects of such a reversed position if we are investigating a process that may be affected by gravity, such as cleaning behaviour. Thus, we determined whether any exuviae or small pathogen-like particles (see §2a) that were placed on downward leaf under-surfaces fell off and confirmed that all exuviae and most small pathogen-like particles that had been placed on leaf under-surfaces did not fall off, even when the surface was facing downward (electronic supplementary material).
(a) What behaviours are used to clean a nest space?
In a previous study, six typical behaviours were observed in S. longus females in their nests : (i) silk weaving, (ii) rest (including feeding), (iii) patrolling, (iv) wiggling, (v) defaecation and (vi) oviposition. Among these behaviours, (i), (iii) and (iv) were highlighted as candidate female behaviours that may be related to nest space cleaning, because we often observed that numerous exuviae were attached to the inner side of a woven nest roof (ceiling) as shown in figure 1, though the mites never moult on the ceiling.
We first carried out an experiment to examine whether an S. longus female cleans its living space (leaf surface) when establishing its nest. We sprinkled globe-shaped silica particles (approx. 20 µm in diameter, scastara-red; Corefront Corporation; hereafter simply referred to as particles) on the leaf surfaces (approx. 70 particles mm−2); thereafter, we individually introduced fertilized females and allowed them to build new nests there. The adopted particle size of 20 µm in diameter is similar to the common size of pathogen spores (spore sizes vary from 1 to 40 µm along the major axis among insect pathogen species ) and near the smallest size that can be observed under a dissecting microscope. After the females started to build their nests, their behaviours were recorded for 12 h under continuous light conditions (approx. 800 fc) using digital video cameras (Panasonic AG-DTL 1H with an Ikegami ICD-878 camera; hereafter referred to as VR). The number of particles adhered to the nest web ceiling was counted at 3, 6, 9 and 12 h after the female began to make its nest web (the video recording was paused at each time point for the 15–30 min necessary to observe the particles). If particles were picked up by certain behaviours, a correspondence between time allocation (and also the number of silk-weaving behaviours) and the number of particles on the web ceiling would be observed. Then, the time allocated to performing each of the (i)–(vi) behaviours and the numbers of silk-weaving behaviours performed by the females during these intervals were counted using the recorded video tape.
(b) Do mites clean nest by silk-weaving behaviour?
Gravid females of S. longus at 3 days after maturity (mated) were individually allowed to construct nests on S. senanensis leaves. Three days after their introduction, we ripped a hole in the middle of each nest web and sprinkled a number of particles through the hole on the leaf surface within the nest.
The working hypothesis of this experiment was as follows: if mites clean their nest space using a silk-weaving behaviour, the frequency of the behaviour should increase with the number of particles sprinkled in nests; but if mites only repair ripped nests, the frequency of the behaviour should not increase with the number of particles. We counted the initial number of the sprinkled particles on the leaf surface inside the nests (n = 18). The number of particles adhered to (beneath) the nest ceiling 24 h after the treatment was then counted. Simultaneously, we recorded mite behaviours for 24 h by VR and counted the number of silk-weaving behaviours the female performed during the period using the recorded video tape.
As a control for the above experiment, we recorded female behaviour in nests with a hole (n = 9) and in nests without a hole (n = 9) by VR and counted how many silk-weaving behaviours the female performed during the 24 h period using the recorded video tape.
(c) How long do silk threads remain adhesive?
Our observations of silk-weaving behaviour suggested that silk threads pulled out by females were often attached to particles, but those making up the web nest were not adhesive. Thus, we inferred that the silk thread quality changes with time. Then we attempted to determine how long the silk threads produced by S. longus females retain their adhesiveness. Twenty-five females of S. longus were individually introduced onto new leaf arenas without nests and particles. We allowed them to spin threads for 5 min and then removed the females. We sprinkled particles (approx. 70 mm−2) onto the leaf areas at different times: 5 min (n = 7), 1 h (n = 6), 3 h (n = 6) and 24 h (n = 6) after the removal of females. Then we counted how many particles stuck to the silk threads after tapping the leaf surface with the shaft of a fine-tipped brush.
(d) Why are the eggs not cleared?
If the silk threads produced by females frequently move the particles from the leaf surface to the ceiling of the nest web (see §3), we wondered why eggs deposited on the leaf surface inside the nest were seldom swept up by the threads (figure 1). We hypothesized that there are mechanisms that prevent S. longus eggs from adhering to the nest ceiling. One plausible mechanism was thought to be wiggling (behaviour (iv), i.e. the female behaviour on the leaf surface just before and after oviposition).
To confirm the above hypothesis, the gravid females of S. longus were individually introduced onto new leaf arenas and allowed to build nest webs. Three days after the beginning of nest web building, we made a tear in the approximate middle of each nest web (31 nests in total). Then the eggs in 16 nests were moved slightly using a fine-tipped pin from the place where they were originally deposited. This treatment was conducted to deactivate possible maternal care of eggs. For the control, no manipulation was made to the eggs in 15 nests.
(e) Does cleaning improve mite survival in the field?
The above four experiments indicated that S. longus females perform nest-cleaning behaviour using silk threads (§3). Therefore, whether such behaviour also functions in their natural habitat is an important question. However, it was difficult to directly evaluate the effect of nest-cleaning behaviour on offspring survival in the field. Therefore, we observed the effect of female attendance under predator-free conditions on the survival rates of the young in nests.
The field experiments were conducted in August 2008 in a small wooded area of deciduous trees with a sub-canopy of S. senanensis on the campus of Hokkaido University. We first arbitrarily selected 33 S. longus nests, each with a single female and 4–11 eggs on the leaves of S. senanensis. Leaf areas of approximately 7 cm2 with single nests were surrounded with a sticky substance called tangle-foot (Fuji Tangle, Fuji Yakuhin Kogyo Inc.) to repel all ambulatory predators (this adhesive substance prevents predators from accessing prey nests ). We employed two treatments as follows: treatment 1 (female attending nest, n = 16), in which we tapped nest roofs with a fine brush from outside but did not remove the female; and treatment 2 (female removal from nest, n = 17), in which we tapped the nest web roofs with a fine brush from outside, drove the female away from the nest and then removed the female. We recorded the number of eggs in each nest at the beginning of the observation period. Two weeks after the treatments, we evaluated the survival rates of the young. It was difficult to evaluate survival rates of the young in treatment 1 because females living in the nests laid additional eggs during the two weeks. Because the eggs deposited at the beginning had developed to at least the protonymph stage over the two weeks, the surviving immatures that had developed from the eggs present at the beginning of treatment 1 should be considered the surviving protonymphs or more developed stages. Under this assumption, we calculated the immature survival rates in treatment 1 as the number of surviving immatures (protonymphs and more developed stages) after two weeks divided by the number of eggs at the beginning of the experiment.
(a) What represents cleaning behaviour?
When introduced onto new leaf arenas, each S. longus female pressed its mouthparts (pedipalpi) onto one side of a depression on the under-surface of the host leaf (S. senanensis) and produced silk threads. The female then turned and walked straight to the other side of the depression (threads are extruded by the walking action ), where she pressed her silk production organs to the leaf once more. The female repeated this process many times. This series of behaviours corresponded to a single nest-building trial , which we refer to as ‘silk-weaving behaviour’ (more specifically, spider mites simultaneously produce two threads that usually unite from their paired silk-production organs [19,20]).
The S. longus females spent a considerable amount of time (up to 3 h) producing new nest webs during the early phases of nest construction. When the female engaged in more silk-weaving behaviour, she picked up more particles off the leaf surface (table 1). As a result, the leaf surface covered by the web nest became cleaner (i.e. the number of particles decreased, as shown in figure 2). As most females rarely performed oviposition, wiggling or patrolling behaviours during the 3 h of nest construction period (table 1), it was apparent that the particles became attached to the woven nest roof (ceiling) of the web nests through silk-weaving behaviours. These behaviours were also performed if the leaf surface was facing downwards (Y. Saito 2010, personal observation).
(b) Is this really cleaning?
However, doubt remained whether the mites perform the silk-weaving behaviour specifically for cleaning the nest space or only for building the nest. If the latter, the cleaning effect is simply a by-product. In well-established nests with an introduced hole, the number of particles attached to the ceiling of the web nest increased with the number of particles initially sprinkled on the nest floor (r = 0.789, p < 0.001; figure 3). The number of particles attached to the ceiling of the web nest also increased with the number of silk-weaving behaviours (r = 0.773, p = 0.0002; figure 4). These results indicate that mites adjust their silk-weaving behaviours in response to the amount of dust in their nest space.
In the control experiments, a female performed an average of 386 weaving behaviours (= number of silk threads spun) in the nests with no particles and no hole. These were considered spinning behaviours in non-treated nests. Since approximately 496 weaving behaviours, on average, were observed in the nests with a hole and without particles, we inferred that 110 silk-weaving behaviours were performed to repair the nest hole (there was a significant difference between ‘no hole’ and ‘hole’ in the web treatments; d.f. = 1, F ratio = 21.25, p = 0.0003 by ANOVA).
As shown in figure 3, we calculated the regression line of the number of silk-weaving behaviours (y) on the number of particles initially sprinkled (x) as y = 0.78x + 510.2 (r2 = 0.622; intercept, p < 0.0001; slope, p < 0.0001). From this line, we obtained an intercept of y = 510.2 when x = 0. This value corresponded well to the value of 496 when a hole was made in the nest web, indicating that the surplus silk-weaving behaviours observed under particle-sprinkling conditions were caused by the existence of particles in nests. That is, more weaving behaviours were observed in the presence of particles.
(c) How long are silk threads adhesive?
The threads produced by females were adhesive just after their production, but these threads became less adhesive with time. As shown in figure 5, approximately 20 particles stuck to the threads when the particles were sprinkled 5 min and 1 h after the production of threads, but there were very few particles stuck to the web when the particles were sprinkled 3 and 24 h after the production of threads. There were significant differences between 5 min versus 3 h, 5 min versus 24 h, 1 h versus 3 h and 1 h versus 24 h, but insignificant differences between 5 min versus 1 h and 3 h versus 24 h according to Scheffe's test (figure 5).
(d) Why are the eggs not cleared?
The eggs of S. longus left intact in the nest never attached to the ceiling of the web nests, and those moved from the original depositing sites were almost all attached to the ceiling of web nests (figure 6). These observations suggest that there are some mechanisms that prevent eggs from being lifted up by the adhesive threads produced by females. We then observed the eggs deposited on the leaf surface within a nest using a scanning electron microscope (SEM) and revealed that the eggs were deposited on the web with a loose weave (figure 7; we call this web a ‘web mat’ hereafter).
(e) Does cleaning increase mite survival in the field?
The immature survival rate in the female attending the nest (treatment 1 in figure 8) was significantly higher than female removed from the nest (treatment 2 in figure 8). The survival rate in treatment 2 after two weeks was 50 per cent, but it was over 90 per cent in treatment 1 (treatments 1 versus 2: p=0.0008 on the arcsine-root transformed values according to Mosteller & Youtz ). Thus, female attendance significantly improved the survival of offspring even under predator-free conditions in the field.
Separating the nest-cleaning behaviour from nest-reinforcement/repairing behaviour has been difficult because these are both commonly actualized by silk-weaving behaviour. We separated these using artificial dust (silica particles) and nest manipulations. The results of our first experiment (§2a) indicated that S. longus females clean the leaf surface through silk-weaving behaviour when they establish their new nests. In our next experiment (§2b), silk-weaving behaviour was actually shown to be enhanced when the nest space is dirty (many particles). These results stress the reality of nest-space-cleaning behaviour in S. longus. If web-spinning behaviour could be performed only for nest repair and enforcement, there would be no need for females to increase the frequency of such a behaviour when the number of particles in their nest increase. Therefore, we conclude that the nest space cleaning is not a by-product of nest repairing and/or reinforcement behaviour. It is apparent that almost all exuviae attached to the ceiling of nest web in figure 1 are also caused by such a nest-cleaning behaviour.
The result of our subsequent experiment (§2c) is very suggestive. The threads spun by the females are adhesive, such that this adhesiveness is believed to function as an adhesive cleaner. The short-termed adhesiveness of threads after production must be advantageous as a material to make solid nest webs. Furthermore, if the threads (web) remained adhesive for a long time, it would be difficult for mites to live within a narrow space covered by it because S. longus need to touch the ceiling of the nest with their dorsal setae to recognize its existence .
The result of next experiment (§2d) showed that some protective means performed to the deposited eggs within nests prevent them from being lifted by silk-weaving behaviour. Based on the SEM observation of S. longus eggs, we found that there was a web mat under each egg (figure 7; note that a web covering eggs is common in spider mites, but a web mat under eggs is extraordinary ). Although direct observation is impossible, this mat seems to be formed by wiggling behaviour always observed just before and after oviposition . Since the silk threads are adhesive 1 h after production, the eggs are fixed in place by the adhesive web mat on the leaf surface. Therefore, we conclude that S. longus females devise such mats to keep their eggs from being lifted up through the silk-weaving behaviour. Several empirical observations suggested that eggs that attached accidentally to the ceiling of web nest (eggs are accidentally kicked and moved by females) were wrapped with dense threads by repeated weaving behaviours and often became difficult to hatch. In other words, the existence of such a mechanism to avoid eggs from the dangerous weaving behaviour strongly indicated that S. longus females perform cleaning of the leaf surface inside the nest. Furthermore, it is suggested that silk-weaving behaviours also target egg surfaces for cleaning because such behaviours must sweep up pathogen-containing dust from the egg surface as well as the leaf surface.
In the field experiment, we observed that the female attendance is effective on the survival of offspring without predation risk. Since many dead eggs and larvae in the non-female-attendance nests were colonized by fungi and/or micro-organism hyphae (though it was difficult to know whether such organisms colonized before or after mite death), one of the factors of their mortality may be pathogen infection . As hypothesized from the laboratory experiments, S. longus females may clean their eggs as well as the leaf surface inside the nest. The positive effects of female attendance under predator-free conditions in the field may be caused by the nest/egg-cleaning behaviour of females, although other factors are not rejected sensu stricto.
Meanwhile, two questions arose from the present results. One question concerns the sources of dust in the densely covered nest space. Stigmaeopsis longus has a special waste-management behaviour: all nest members deposit faeces at a certain place (figure 1 ). This is realized by tactile and chemical cues, and is considered a cooperative nest sanitation mode . However, even if nest members keep to the regulation, this waste-management adaptation may not always be perfect because such a faeces mass outside the nest is easily infested by various fungi and bacteria . All nest members periodically go to the depositing place for defaecation, such that it is inevitable that they will collect some spore and/or micro-organism on their legs and body from the faeces mass. Because they move around the nest space after depositing faeces, there is a great risk to propagate spores/micro-organisms over the nest surface or on eggs. Exuviae with dust are also dangerous for mites because they may carry pathogens, such that they must be managed by the nest-cleaning behaviour.
The other question concerned why mites attached dust and exuviae to the ceiling of the nest web. Fungus spore and bacteria attached on the ceiling of the web nest are thought to easily dry up and become harmless in comparison with the leaf surface. Furthermore, repeated silk-weaving behaviour must wrap them within the woven roof. Then the danger of infection for mites living under the web seems to be well mitigated.
The discovery of communal sociality in a spider mite  (S. longus, a tiny organism that lives on leaf surfaces and has very simple neural systems) was an early topic in behavioural ecology . Later, the details of sociality, cooperative nest building and cooperative faeces-manipulation behaviours were reported [15,16]. In this study, we uncovered another kind of social behaviour in S. longus. Cleaning the nest space and egg with silk threads is a new mode of social behaviour observed in the social spider mites. Self-produced silk threads are not tools in the strict sense of behavioural ecology, but we learn that S. longus use silk threads as if they were a nest-cleaning ‘device’ like diverting adhesive tapes for sweeping.
Although silk threads have evolved independently in many different arthropods (e.g. mites, spiders, moth larvae, Hymenoptera and Psocoptera [25–27]) and may serve several functions—such as providing nest (cocoon) material, prey-capturing devices and lifelines—the nest-cleaning function observed in S. longus is the first such instance to be reported in the animal kingdom. As implied by their name, most spider mite species produce silk threads that function as basic lifelines for residing in arboreal habitats . Such threads are sometimes used in different ways to disperse from deteriorated habitats to new ones . Furthermore, several species use threads to construct web structures on leaf surfaces to prevent predation and mitigate adverse climatic conditions [6,29]. Here we have demonstrated another extraordinary use of silk threads as a device for nest space and/or egg cleaning. Such diversity in the usage of a single material strongly suggests that silk threads play a crucial role in the adaptive radiation of this taxon, Tetranychidae. Furthermore, we conclude that the cooperative social living observed in S. longus was able to evolve because of their production of silk threads that have the potential to be used in diverse ways, including as devices to allow this species to establish aggregative and sedentary lifestyles.
We thank Anthony R. Chittenden for English editing, and Daigo Aiuchi, Masanori Koike, Takane Sakagami and Yukie Sato for valuable advice. This work was supported by JSPS KAKENHI no. 20370006.
- Received September 5, 2010.
- Accepted October 21, 2010.
- This Journal is © 2010 The Royal Society