Experimental & Applied Acarology, 23 (1 999) 4 Effects of the nest web and female attendance on survival of young in the subsocial spider mite Schizotetranychus longus (Acari: Tetran y chidae) Kotaro Mori*, Yutaka Saito and Takane Sakagami Laboratory of Animal Ecology, Graduate School of Agriculture, Hokkaido University, Sapporo 060-8589, Japan (Received 30 June 1998; accepted 27 September 1998) ABSTRACT The subsocial spider mite Schizotetranychus longus lives gregariously in woven nests on leaves of Sasa bamboo. Adults of both sexes defend their young against the predatory mite Typhlodromus bambusae. The effects of web and female attendance of this species on offspring survival were evaluated in a natural forest. Experimental removal of web and females revealed that S. longus young suffered greater mortality than in the control. Furthermore, the web made by parent females had a positive effect on offspring survival, possibly through preventing predators from intruding into the nest. Attendance of a female in a nest also had an effect on improving the * survival rate of her offspring over a short period. We could not detect any function of the web and female other than protection against predators at least for the 5 day period of the experiment. The nest web of S. longus has an important function in the survival of young by preventing the entry of pedestrian predators (generalist) and females may play a role in defending against specialized predators which can intrude into the nests. Exp Appl Acarol 23: 41 1-418 63 1999 Kluwer Academic Publishers Key words: Schizotetranychus longus, protection, defence, nest web, predator, subsociality. INTRODUCTION Spider mites (Acari, Tetranychidae) infest the leaves of many plant species and usually produce webs over living surfaces. These webs can be categorized according to their shapes and densities. Some webs function as a 'protective refuge' against predators in the laboratory (McMurtry and Johnson, 1966; Takafuji and Chant, 1976). Schizotetranychus longus is a subsocial mite infesting the dwarf bamboo, Sasa senanensis. They build woven nests along the midribs or curled edges of the lower leaf surfaces and live gregariously inside the web with their offspring (Saito, ' * To whom correspondence should be addressed at: Tel: +81-11-706-2491; Fax: +81-11-757-5595; e- mail: kotaro@res.agr.hokudai.ac.jp 01 68-8 162 @ 1999 Kluwer Academic Publishers
412 K. MORI ET AL. 1986a). They extend their nest unit by unit (or cell) when their populations increase and these cells become linked together (Saito and Takahashi, 1982; Saito, 1986a). Its 'life type', defined by using the patterns of silken threads (Saito, 1983, 1985), is a highly developed one (subtype c in the web-nest type, denoted as WN-c) in the Tetranychinae (Saito, 1983). An extremely densely woven roof and a fixed defecation site outside each nest characterize WN-c. Although Saito and Takahashi (1982) and Saito (1983, 1985) expected the nest web of S. longus to be effective, because it seemed dense enough to prevent predators' intrusion, no information is available about how effective it is either in the field or in the laboratory. In addition, Saito (1986a,b) reported that the females and males of S. longus counterattack their specific predatory mite Typhlodromus bambusae, which intrudes into the nests to feed on their offspring. It was experimentally demonstrated that S. longus sometimes killed predatory larvae and nymphs and frequently drove away most stages of the predator from their nests. The effects of this counterattack behaviour have not been evaluated under natural conditions. More than ten species of tetranychid predators including mites and insects were observed on S. senanensis (Saito, 1990). Some may be generalists that cannot intrude into the nest web (K. Mori, unpublished), whereas others may be specialists that can. In this study, we experimentally examined whether the webs and attendance of adults could protect young spider mites from predation under field conditions and also evaluated how effective each of these defences was. Because there were few males in the newly founded nests of 5'. longus used in this experiment, female defence efficiency is important during the early period of foundation. We evaluated only the effect of maternal attendance. MATERIALS AND METHODS Experiments were conducted in a deciduous forest on the campus of Hokkaido University, Sapporo, Japan, from July to August 1996. The main vegetation of the lower layer of the forest consists of 5'. senanensis. From late June to early July in Sapporo, the plant grows new shoots and begins to develop new leaves. The leaves are evergreen and perennial and can survive for up to 3 years. During this period, S. longus females tend to disperse from their natal nests on old leaves and colonize young leaves to found new colonies. Newly established nests usually include foundresses and their immatures and are very easy to manipulate. Therefore, we used colonies of S. longus on new leaves for the present experiment. To evaluate the effectiveness of the nest web and/or female attendance in defending young against predators, we performed six experimental treatments on S. longus nests: female removal (n = 191, female and web removal (n = 25), female removal with tanglefoot (a sticky substance named 'Fuji tanguru', supplied by Fuji
EFFECTS ON SURVIVAL OF YOUNG S. LONGUS Yakuhin Kogyo Company) (n = 20), female and web removal with tanglefoot (n = 21), tanglefoot treatment (n = 15) and control (no treatment and a single female in a nest, n = 21). We looked for newly founded nests on new leaves that consisted of one to two nest cells including a single female and her offspring (three to 24 eggs and/or larvae). Next we carried out several combinations of the following three treatments on randomly selected nests: (1) carefully removing nest webs with a fine needle (for treatments of female and web removal and female and web removal with tanglefoot), (2) driving out females from nests by tapping the outer surfaces of the web roofs with a fine brush (for female removal and female removal with tanglefoot) and (3) encircling nests with tanglefoot to exclude walking predators (for female removal with tanglefoot, female and web removal with tanglefoot and tanglefoot treatment). We could not set up the web removal and female attendance (with tanglefoot) experiment (web removal and web removal with tanglefoot) because S. longus females never attend their young if there is no web present. We then counted immatures (eggs and larvae) in each nest and marked leaf surfaces beside the nests with India ink (the number of individuals recorded here is denoted as NI). Five days later, we again recorded the number of immatures (NA) and calculated the survival rate over 5 days for each nest (NAINI). An experimental duration of 5 days was decided upon because under laboratory conditions (25 2 1Â C 50-60% RH and 15:9 h L:D) (Saito and Ueno, 1979) a period of 5 days is not long enough to allow full development of 5'. longus from egg to adult. As the mean temperature in Sapporo that year was below 25OC, it was possible that the immatures in the marked nests displayed low activity and did not carry out any weaving or defence behaviour. In the tanglefoot treatment and control groups, it was unavoidable that the females in the nests oviposited during the 5 days. We counted both surviving and dead individuals and estimated survival rates as follows. Since the number of surviving immatures in all the nests with tanglefoot treatment increased over 5 days and because no dead bodies were found, we considered that no initial individuals had died in any tanglefoot treatment nests (survival rate = 1). Concerning the control group, there were three different outcomes. i (1) If NA equalled or increased in comparison with N1 and we could not find any dead immatures, the survival rate was estimated to be 1. (2) If NA = 0, the survival rate was estimated to be 0. (3) If NA was not 0 and was less than N1, the survival rate was estimated as NA/ (N1+ NE), where NE (= 2.37) was the estimated number of ovipositions per female during the experimental period. This case was observed only in the female- ' decreased nests (see next paragraph). Because the mean increase in immatures in the tanglefoot treatment nests over 5 days was 4.73 and as we believed that the disappearance of females in these nests occurred at the midway point of the experimental periods, we estimated that the
414 K. MORI ET AL. number of ovipositions in the female-decreased nests was NE. These assumptions may cause some over- and/or underestimation. During the experimental period, it was possible for the number of females per nest in the female removal and control experiments to change due to predation, immigration andor emigration, In fact, these phenomena were only observed in some nests of the control group. We omitted these data from the analyses and will refer to this' problem in the Discussion in relation to the effectiveness of female attendance, Proportion data were transformed using the arcsine square root in the manner of Freeman-Tukey transformation (Mosteller and, Youtz, 1961) when ANOVA and multiple comparisons (Scheffe's method) were applied. To test the effect of the web and female, we performed two-way ANOVA using the StatView package for Macintosh (Abacus Concepts, Inc., Berkeley, California). RESULTS ), mber of females in 11 out of 21 nests of the control group decreased to 0 and those in four nests incriased to 3.5 (mean), while six nests kept the same number (i,e. a single female), Low survival rates in the female-decreased nests (mean = 0.70 radian) greatly influenced the mean survival rates in the control group (1-02, cf. increased nests 1.43 and unchanged nests 1.33). W? thus used the data from unchange'd nests (n = 6) to analyse the effect of female attendance below. Existence of a web and a female affected offspring survival of S. longus (Fig. 1). Experimental groups with tanglefoot (female removal with tanglefoot, female and web removal with tanglefoot and tanglefoot treatment) had higher offspring survival than groups without it (female removal and female and web removal) (Fig. 1). These suggested that predation pressure by pedestrian predators on S, long& is potentially very high. We observed evidence of several predators in the experimental nests, although we did not determine their species. These were a phytoseiid female (in control), a phytoseiid nymph (in female removal), a phytoseiid cast skin, a phytoseiid egg (in female md web removal) and larvae of a gall midge (Cecidomyiidae) (in female removal with tanglefoot and control). To test the effect of the web on pedestrian predators, we performed a two-way ANOVA for no-female groups (female removal, female and web removal, female removal with tanglefoot and female and web removal with tanglefoot) (Table 1). There was a significant web X tanglefoot interaction. The web's effect was detected in non-tanglefoot nests (female removal versus female and web removal; p < 0.001, Scheffe's multiple comparison for no-female groups) but was lost in those with tanglefoot (female removal with tanglefoot versus female and web removal with tanglefoot, NS). Furthermore, the web could not fully protect against pedestrian predators, as could tanglefoot (female removal versus female removal with tanglefoot; p < 0.01).
EFFECTS ON SURVIVAL OF YOUNG S. LONGUS 415 To test the effect of female attendance on pedestrian predators, we performed a two-way ANOVA for groups with the web (female removal, female removal with tanglefoot, tanglefoot treatment and unchanged nests of control) (Table 2). There was a significant female X tanglefoot interaction. The effect of female attendance was detected in non-tanglefoot nests (female removal versus control; p < 0.01, Scheffe's multiple comparison for groups with the web) but was lost in those with, tanglefoot (female removal with tanglefoot versus tanglefoot treatment; NS). Experimental group Pig. 1. Effects of the web and females on offspring survival against predation. Values are mean and SD (radian). Data calculated after Freeman-Tukey transformations for observed values. Treatment groups with different letters are significantly different (p < 0.05, Scheffe's test for all experimental groups).
EFFECTS ON SURVIVAL OF YOUNG S. LONGUS 417 then attendance by a single female has little effect on the survival of young (see Saito, 1986b). Defence is more effective in the laboratory when the nests include many adults of S. longus (Saito, 1986b). Actually, in newly founded nests more than one female was often observed (Saito, 1987). The counterattack of S. longus is expected to be more effective in well-developed nests with many females and males. Weneed to, examine the effects of male attendance and how the number of adults in a nest affects defence efficiency in order to determine the significance of biparental defence and subsociality in S. longus under natural conditions. The life type of a spider mite as a defensive tactics may be related to the structure of the community in which the species takes part (Hardin and Tallamy, 1992; Kudo, 1996). There are many kinds of predators on S. senanensis including predatory mites. These include six phytoseiids (T bambusae, Amblyseius ainu, Amblyseius orientalis, Amblyseius womersleyi, Phytoseius tenuiformis and Phytoseius sp.), two stigmaeids (Agistemus summersi and Agistemus iburiensis), a tarsonemid, a cunaxid and insects (a thripid, a staphylinid, a cecidomyiid and a coccinellid) (Saito, 1990). These predators include both generalists and specialists. For example, A. womersteyi and A. orientatis occur on many plants with various spider mite species and may be generalists. T bambusae and A. iburiensis, on the other hand, are known as specialists of the Schizotetranychus celarius species complex (including S. longus), because they occur exclusively inside the nests of this group in Japan (Takahashi, 1987; Saito, 1990). The nest web of S. longus might have evolved as a defensive refuge against such generalist predators. On the other hand, the finding that the web could not always exclude all predators suggests that S. longus still faces the risk of predation by specialized predators, such as T bambusae, I? tenuiformis and A. iburiensis, which easily intrude through small openings in the nest without destroying webs (Saito, 1986b; K. Mori, unpublished). These predators may be specialized in their shape (body thickness, dorsal hairs and so on; see Sabelis and Bakker, 1992) and/or, behaviour for invading S. longus nests for predation. The counterattack behaviour by S. longus parents (males and females) and subsociality may thus have evolved through interactions with such specialist predators, after evolving its nest web, which provides an apparently distinguishable space for defence. The S. celarius species complex on bamboo contains at least four types (one is S. longus and the other three are S. celarius) making various sizes of nest cells (Saito and Takahashi, 1982; Takahashi, 1987). Among them, S. longus has the largest nest cell (approximately 10 rnrn2) and a defensive behaviour that has not been observed in the other three types. This suggests that there are close relationships between nest size and biparental defence in this species complex. We think that the predator-prey interaction may play an important role in the evolution and maintenance of their life types. These hypotheses must be tested by mapping the characteristics of the four types, e.g. group sizes, nesting patterns and behaviour, onto their phylogeny, as these may be related to the tactics they use to avoid predators.