OUP user menu

Conditional Mutualism Between Allodynerus delphinalis (Hymenoptera: Vespidae) and Ensliniella parasitica (Astigmata: Winterschmidtiidae) May Determine Maximum Parasitic Mite Infestation

K. Okabe, S. Makino
DOI: http://dx.doi.org/10.1603/EN09208 424-429 First published online: 1 April 2010


Mutualism is a prominent interaction within ecosystems, yet most may actually be conditional. The symbiotic mite, Ensliniella parasitica Vitzthum, ingests the hemolymph of juvenile potter wasps, Allodynerus delphinalis (Giraud), but also protects them from a natural enemy, Melittobia acasta, and is transported to new nests in host pockets specialized for this purpose (i.e., acarinaria). Thus, two different antagonisms from the mite may arise: commensalistic cheating is expected without the natural enemy of the host, and parasitism is predicted with excessive numbers of the mite. However, facultative parasitism mediated by mutualism has rarely been studied in any organism. We found no significant differences in juvenile mortality, nesting rate, or fecundity between mite-free and naturally mite-laden juveniles. However, when overloaded with mites (≈1.5–2.5 times more mites than the maximum number per wasp larva in the field), the developmental period of the male wasp was significantly delayed, and juvenile wasp mortality increased to 30%. These results show that mutualism mediated by parasitism may revert to parasitism, suggesting that either or both organisms in a mutualism mediated by parasitism need population control of the parasite to avoid the risk of parasitism.

  • acarinarium
  • Eumeninae
  • guarding mutualism
  • host control
  • parasite

Mutualism is a prominent interaction within various ecosystems, yet mutualisms are often unstable and highly conditional, because of the unpredictability of biotic and abiotic environments (Bronstein 1994, Herre et al. 1999, Kersch and Fonseca 2005, Karst et al. 2008). For instance, ant protection to plants or homopterans weakens with high natural enemy density and/or cheating in the absence of enemies (Cushman and Whitham 1989, Rico-Gray and Oliveira 2007). The cleaning mutualism between gobies and longfin damselfish is also temporal, shifting between mutualism and parasitism (Cheney and Côté 2005). Evolution can also cause the interaction to change from parasitism to mutualism and vice versa, as seen in microorganisms such as ectomycorrhizal fungi (Karst et al. 2008) and Wolbachia microbes (Pannebakker et al. 2007) or from various antagonisms to mutualism, as in ant-plant systems (Rico-Gray and Oliveira 2007).

Some bees and wasps associated with mites have a peculiar structure called an acarinarium, which functions solely to transport phoretic mites to a new nest (OConnor and Klompen 1999). The relationship between acarinarium holders and users has been fully elucidated that most cases are thought to be mutualistic (summarized by Eickwort 1994), but only a few cases are suspected to be antagonistic (Klimov et al. 2007). For instance, the transport of mites in the acarinaria of Mesotrichia, Koptortosoma, Platynopoda, and Afraxylocopa carpenter bees is most likely mutualistic, with the mites perhaps removing harmful microbes from juvenile hosts while feeding on host exudates; in this interaction, neither positive nor negative effects on host development have been observed (Skaife 1952, Madel 1975, Eickwort 1994). Okabe and Makino (2008b) for the first time reported the evidence of mutualism mediated by acarinaria between the potter wasp, Allodynerus delphinalis, and its host-specific symbiont, Ensliniella parasitica. They showed that the mite protected wasp juveniles from the parasitoid, Melittobia acasta (Walker), by continuous mass attack probably harming membranous parts of the parasitoid to death of either of them when the parasitoid attacks the prepupal or pupal wasp in a cell. However, the relationship between A. delphinalis and E. parasitica is also conditional: under parasitoid-free conditions, the mite is parasitic, feeding on the hemolymph of juvenile wasps, although normally without negative effects on host development and thus behaving as a commensal (Okabe and Makino 2008a). We suggest that, under the condition with the natural enemy, both wasps and mites come to an agreement to maintain mite numbers large enough to kill the natural enemy to protect juvenile wasps (for the mite, as a food resource). Furthermore, from the phoretic mite point of view, the mite should invade a host cell as soon as possible to avoid being involved in the death of the adult host. Therefore, with or without the parasitoid, numbers of the mite invade into cells made earlier. For the wasp, however, the mite become cheater when no natural enemy (no-guarding) or too many mites per cell might hart its offspring by overexploitation, i.e., excess intake of wasp juvenile hemolymph (parasitism). This raises the question: does the wasp or perhaps the mite actually need to control mite numbers to avoid mite parasitism?

Along the mutualism-parasitism continuum, an interaction might go over to cheating without a reward or to overexploitation. Although it may be difficult to determine the outcome in either case, cheating has frequently been examined (e.g., yucca and yucca moth, Segraves et al. 2008; cleaning goby, Cheney and Côté 2005), but overexploitation has scarcely been studied. We examined the negative effect(s) on the juvenile host of mite overload (i.e., overexploitation of the wasp juvenile by the mite), defined as the number of mites per cell beyond the maximum (23), as observed in the field and based on the work of Okabe and Makino (2008a). We also compared the effects of a natural load and mite-free conditions on juvenile survival, adult nesting rate, and fecundity of infested and noninfested wasp juveniles. Based on the results, we suggest that acarinaria may have developed to maintain the parasitic mite as a mutualist.

Materials and Methods

Wasps and Mites.

The host wasp A. delphinalis is distributed from Europe to Japan (Klompen et al. 1987, Yamane 1990) and nests in dead stems of Rubus spp. (Rosaceae), Sambucus racemosa (Caprifoliaceae), Hibiscus moscheutos (Malvaceae), Solidago canadensis, Bidens frondosa, Erigeron canadensis, and Conyza sumatrensis (Compositae), excavating the pith in the middle of a stem (Enslin 1922, Benno 1945, Crèvecoeur 1945, Okabe and Makino 2008a,b). The wasp provisions a maximum of seven brood cells (≈4.5 mm diameter, 20 mm long) per nest (Enslin 1922). Its life cycle is similar to that of other tube-renting eumenine wasps: the adult female lays an egg in a brood cell stocked with paralyzed lepidopteran prey; a hatched larva develops by feeding on the prey, pupates after a prepupal period lasting from 1 wk to the entire winter, and eventually emerges as an adult (Okabe and Makino 2008a).

The mite E. parasitica is specifically parasitic to A. delphinalis (Klompen and OConnor 1995). Dispersal deutonymphs are housed in host acarinaria during phoresy (OConnor and Klompen 1999, Makino and Okabe 2003). The mite life cycle has been studied in detail (Okabe and Makino 2008a): the deutonymphal mite invades a cell during host oviposition and molts to the tritonymph and quickly to the adult stage, taking host and host prey hemolymph; the female mates with a large or small male (i.e., her parthenogenetically reproduced son) and begins laying eggs on the host at host pupation; larvae and protonymphs acquire nutrition from the pupa and develop into deutonymphs around host eclosion.

Voucher specimens of this study are deposited at the Forestry and Forest Products Research Institute.

Mite Sampling and Rearing.

We collected A. delphinalis nests in several grasslands dominated by S. canadensis and C. sumatrensis in Tsukuba, Ibaraki Prefecture, Japan, between 2004 and 2007 for nest examination and rearing experiments. In the laboratory, we opened dead stems containing wasp nests and recorded nest contents. We maintained closed nests with juvenile wasps on cotton sheets spread on the bottom of plastic containers (75 by 160 by 12 mm) at ambient temperature.

The rearing method followed that of Okabe and Makino (2008a,b): a mated female was transferred to a cage (450 by 450 by 450 mm), to which we supplied dead S. canadensis stems (20 cm long) as nest material and 10-20 immature Brachmia triannulella macroscopa (Lepidoptera: Gelechiidae) with its food plants (Ipomoea batatas and Calystegia japonica, Convolvulaceae) every day as prey. Soon after a mother wasp completed a nest, we collected the nest and observed the cell contents.

For the experiments, we used both field- and laboratory-reared juveniles maintained at outdoor temperatures. For the overload experiment, we maintained nests at 25°C to compare the exact developmental periods of over- and naturally loaded cells to mite-free cells. The mortality of mite-laden and mite-free juvenile wasps was compared in both field-collected and laboratory-reared nests after visible parasitism, except when mites were excluded. For nesting rate and fecundity, we defined a mite-free wasp as one that had never been infested by mites since the egg stage. Fecundity (number of eggs per female) and nesting rate (number of females nesting/total number of females) were examined only in the rearing experiment. We also recorded the duration of the prenesting period. The experiments were terminated when wasps died or did not make a nest for 3 wk under identical rearing conditions. For the fecundity assessment, non-nesting females were excluded. Nesting ratio was calculated as total female numbers engaging in nesting (at least, excavation of a dead stem) divided by the total female numbers reared for nesting.

Overload Experiment.

Mite numbers ranged from 1 to 23, the mode of mites per cell was 5, and the average was 6.45 ± 4.3 (SD; N = 348) in the mite-infested field nests (Okabe and Makino 2008b), whereas they ranged from 1 to 21, and average numbers of mites were 5.79 ± 4.0 (N = 311) in the mite-infested laboratory nests (Okabe and Makino 2008a and this study). We manipulated mite numbers per cell to ≈50 (almost double of unmanipulated mite numbers) by adding 30-50 mites to a cell containing a wasp egg. The nest was maintained in its plastic container at 25°C until the juvenile became an adult or died. As a control, mites were excluded from cells containing wasp eggs, reducing the mite number per cell to 0-3; these were maintained at 25°C. To add and exclude mites, we carefully removed anesthetized moth larvae by forceps from a cell, checked mite numbers in the cell, and collected mites with a faint painting brush a little moisten with distilled water. After the manipulation, we carefully packed the moth larvae back in the cell. The wasp developmental period and mortality were checked every 24 h.

Statistical Analyses.

We used χ2 tests to analyze mortality differences between mite-free and either naturally loaded or overloaded juveniles reared in the laboratory. Mite-free and mite-laden female nesting ratios were compared using χ2 tests, and fecundity was analyzed using t-tests, following the Shapiro-Wilk W-test for normality. After comparing developmental periods between overloaded and underloaded juveniles using the Mann-Whitney U test, we studied whether the developmental period was related to mite number. Because the developmental periods of female and male juveniles differed significantly (Mann-Whitney U test P = 0.000334, n = 11 females, 12 males), the developmental period of each sex was compared separately. We used the statistical package STATISTICA 06J (StatSoft 2005) for all statistical analyses.


Unmanipulated Samples.

The mortality of juvenile hosts in naturally mite-laden or mite-free cells did not differ in either field-collected or laboratory-reared nests without predators or parasitoids (Table 1). Eight wasps failed to complete ecdysis, one did not hatch, and the others died for unknown reasons, but probably because of physiological problems, diseases, or inappropriate food resources except one: a male pupa with 21 mites failed ecdysis. The other juvenile with 20 mites was female and did not show any difference in development. Excessive mite parasitism did not seem to kill the host.

View this table:
Table 1.

Fecundity did not differ significantly between mite-laden and mite-free female wasps reared in the laboratory (mean ± SD = 12.00 ± 5.5, n = 26, and 8.75 ± 2.5, n = 8, respectively; t-test, P = 0.11154; after the Shapiro-Wilk W-test for normality, P = 0.0877; Fig. 1). The average egg number of reared females with and without mites was 11.24 ± 5.1 (n = 34). Nesting ratios did not differ significantly (P = 0.2989 by χ2 test) between females that were mite-laden (83.3%, n = 36) and those that were mite-free (70.0%, n = 10) as juveniles. The time between release into a rearing cage 7 d after eclosion and the start of nesting, as determined by plant stem excavations, was 8.9 ± 5 d in infested wasps (n = 26) and 9.8 ± 6 d in mite-free wasps (n = 4), but these values did not differ significantly (Mann-Whitney U test, P = 0.69169).

Fig. 1.

Fecundity of mite-laden and mite-free A. delphinalis. Mite-laden females were parasitized by the mite E. parasitica, whereas mite-free females were raised without mites,

Effect of Mite Overload.

At manipulation, the number of mites on a prepupal host ranged from 32 to 52 (mean ± SD = 42 ± 5.8). Although some disappeared for unknown reasons, most mites (at least 30 per sample) survived on the host until host eclosion or death. Those on living females and males numbered 41.3 ± 6.0 (n = 6) and 43.0 ± 6.2 (n = 4), respectively, with no significant difference between the sexes (Mann-Whitney U test, P = 0.8311). The developmental periods of overloaded and less-loaded (the average mite number was 2.75 ± 1.7) females were 21.7 ± 1.4 (n = 6) and 21.0 ± 0.7 (n = 11) d, respectively, with no difference between the groups (Mann-Whitney U test, P = 0.99343). The developmental periods of overloaded and less-loaded males were significantly different (Mann-Whitney U test, P = 0.005285) at 20.5 ± 0.6 (n = 4) and 17.75 ± 1.3 (n = 12) d, respectively. Whereas developmental period was not correlated with mite infestation numbers in female wasps (P = 0.7086; Fig. 2, top), the correlation was significant for male wasps (P = 0.002; Fig. 2, bottom).

Fig. 2.

Developmental periods of female (top) and male (bottom) A. delphinalis at 25°C. Mite numbers were maintained between 0 and 49 at the wasp egg stage in female and between 0 and 52 in male.

Mortality in unmanipulated laboratory nests was 12-15% but increased to 30% in nests overloaded with mites (Table 1). Although the causes of death were uncertain, we found no evidence of parasites or fungal disease except one male death with 21 mites. Feeding marks on overloaded hosts were conspicuous. One dead host was in the final instar, with 48 adult mites; two hosts were in the prepupal stage, with 28 and 39 mites, respectively; and one male was in the early pupal stage, with 55 mites. We also found a female wasp dead at eclosion, with 32 adult mites and a large number of juveniles. Because sexes of wasps cannot be identified until pupa, sexes of dead wasp juveniles were unknown except two died in the pupal stage.

Deaths of juvenile wasps with few or no mites occurred at any stage between the early instars and eclosion for unknown reasons. Although most (37%) deaths occurred at the prepupal stage with an average number of mites (around six mites), one host died in a mite-free cell.


Our results showed for the first time a shift between cheating (no guarding) and overexploitation (parasitism) in a guarding mutualism mediated by parasitism. In the laboratory experiment without added mites, juvenile mortality, fecundity, and nesting performance did not differ between wasps that were mite-laden or mite-free during the juvenile stage, although a male infested with 21 mites failed to complete ecdysis (Tables 1; Fig. 1). However, when mites were added to a cell at levels above the field maximum, juvenile deaths occurred more often. The development of female juveniles was not affected by mite overload, probably because females are larger than males and thus less sensitive to parasites (Yamane 1990). In contrast, a prolonged male developmental period was significantly correlated with mite infestation numbers (Fig. 2). We suspect that it occurred in unmanipulated samples, but because of the small sample numbers of mite overload, a statistically significant difference was not shown. Furthermore, although differences between the sexes are unknown, juvenile mortality rose to 30% when overloaded with mites (Table 1). Tube-renting bees and wasps provision a linear sequence of cells in preexisting cavities, including former insect galleries (O'Neill 2001). Within a nest, A. delphinalis female cells are established interior to male cells so that the female, who spends a longer time in development, can emerge from the nest after the males (unpublished data). A. delphinalis completes a maximum of one cell per day and makes at most seven cells in a nest (Enslin 1922, Crévecoeur 1945, Okabe and Makino 2008a). Therefore, the male life history traits of fast development and nest exodus soon after eclosion are necessary if males in exterior cells were heavily infested with mites. A dead individual also plugs the nest tunnel, although mites can migrate onto a phoretic host from another cell in the same nest (Okabe and Makino 2003).

Neither the host wasp nor the parasitic mite could expect to fully benefit from a very small number of mites per cell, although tolerating a small number of mites (one or two) may be a better strategy than no mites because both the mite and wasp offspring could inherit numbers of mites if both were able to avoid parasitoid attack given that the mite has female-biased sex ration and a virgin mite female can reproduce offspring after mating with a parthenogenetically reproduced son (Okabe and Makino 2008a). Harboring a large number of mites in a cell does not seem to benefit either the wasp or the mite: although having many mites could ensure a higher probability of survival, the interests of the wasp and mite are incompatible if excessive mites become parasite and/or if most mites invade the earlier provisioned cells and latter cells have to survive without guard. Thus, the wasp must develop control measures to maintain a reliable mite population to protect its offspring from a natural enemy; the mite need not confront the measure, because it could rely on their mutual need for mites in each cell, unless excessive mites kill the host.

In mutualism mediated by antagonistic interaction such as parasitism, there is a risk that an originally antagonistic partner might cheat and go back in antagonist. In such case, the other has to develop control measures against the cheater so that they could keep relationship to take reciprocal benefit. In a legume-rhizobium mutualism, for instance, host sanctions by the legume were observed when the bacteria took nutrients from the host but did not gave back fixed nitrogen from the air (Kiers et al. 2003). In an ant-plant mutualism, retaliatory sanctions of degradations of plant domatia (a certain plant structure providing shelters for symbionts) were applied against nonpatrolling ants (Edwards et al. 2006). Among the enslinielline mites, close relatives of EnsliniellaKurosaia, Kenethiella, Stenodynerus, and its sister group Vespacarus—represent parasitism to their specific hosts (Krombein 1967, Cowan 1984, Okabe and Makino 2003). It may suggest that mutualism seen in potter wasps and ensliniellines arose from mite parasitism. Thus far, we have not observed mite self-regulation, but acarinaria may control harmful mites, as seen with the mite pockets developed by Gecko lizards for chiggers and the acarinaria in bee-mite relationships that minimize the negative effects of the mites (Arnold 1986, Benton 1987, Klimov et al. 2007).

We suspect that the parasitism of E. parasitica is maintained at an undetectable level under natural conditions except a very few cases because of self- or host-control. The metasomal acarinarium of A. delphinalis is sexually dimorphic: males have a simple depression on the second tergite, whereas females have a deep depression covered by the exoskeleton with a small opening on each side, suggesting that the openings for mites have become smaller during the evolutionary process (Giordani Soika 1985, Makino and Okabe 2003). Given that only one mite at a time can go through the opening and that mites invade wasp cells only during wasp oviposition (Okabe and Makino 2008a), a small opening might control the number of mites entering and exiting the acarinarium. Because acarinaria on bees and wasps are structures that harbor mites in the phoretic (nonparasitic) stage, the use of an acarinarium to stop mite transmission is almost impossible. Better strategies to exclude antagonistic mites would be to have no acarinarium (i.e., carry fewer mites), as seen in many bees and wasps associated with mites, or to remove inquiline mites from a nest, similar to Ancistrocerus antilope (Cowan 1984). Thus, we hypothesize that the acarinarium originally evolved to harbor useful mites and then the wasp also uses it to control the number of invasive (i.e., potentially harmful) mites entering a cell.


This study was supported by a Grant-in-Aid for Scientific Research (C), 2006, 185800560001 from the Japan Society for the Promotion of Science.

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/), which permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions{at}oup.com


View Abstract