- Email: [email protected]
Soldiers effectively defend aphid colonies against predators in the field
Anim. Behav., 1998, 55, 761–765
Soldiers effectively defend aphid colonies against predators in the field WILLIAM A. FOSTER & PHILIP K. RHODEN Department of Zoology, University of Cambridge (Received 22 April 1997; initial acceptance 20 June 1997; final acceptance 28 August 1997; MS. number: 5530)
Abstract. Morphologically specialized soldiers occur in more than 50 aphid species in the families Pemphigidae and Hormaphididae. To study the effectiveness of soldiers of the gall-forming aphid, Pemphigus spyrothecae Pass., in protecting their galls against natural levels of predation, we manipulated the proportions of soldiers and non-soldiers in sets of galls still attached to poplar trees in the field. Galls with 50 soldiers and 50 non-soldiers were approximately 10 times less likely to be attacked by predators than galls that contained 100 non-soldier aphids. There were significantly fewer live aphids, and significantly more dead aphids, in galls without soldiers than in galls protected either by soldiers or by being within a bag. There were no significant differences in the survival of aphids in galls protected by soldiers compared with those protected by bagging. The soldiers did not protect the galls against invasion by the cohabiting aphid Chaitophorus leucomelas Koch. These observations provide the first demonstration that soldiers are effective in defence against natural levels of predation under field ? 1998 The Association for the Study of Animal Behaviour conditions. Communal defence against predators and parasites is one of the most commonly invoked factors favouring the evolution of cooperative behaviour in animals (e.g. Alexander et al. 1991; Bourke 1997). However, morphologically specialized warrior castes are rare in social animals and are found only in termites, several genera of ants and two families of aphids (Choe & Crespi 1997). There are very few experimental studies of the effectiveness of soldiers in any of these animals, although it is generally accepted that the highly specialized soldiers in ants and termites can effectively protect their colonies. Because aphids are generally perceived as being weak and defenceless, and also because the soldiers (unlike those of ants and termites) are first instars and therefore the smallest possible life-history stage, it has been difficult to convince people that they are actually capable of effectively defending their colonies. Aoki (1977), in the original description of a soldier aphid, Colophina clematis, did show that soldiers could kill predators and there have been further laboratory demonstrations since then (summarized by Stern & Foster 1996). Correspondence: W. A. Foster, Department of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, U.K. (email: [email protected]). 0003–3472/98/030761+05 $25.00/0/ar970664
Our aim in this paper is to demonstrate experimentally that aphid soldiers can effectively protect their colonies against predation under field conditions. It is important to establish this point, since all previous experimental studies of predation of soldier-producing aphids have either used predators that are not natural predators of the particular species or have presented natural predators to the galls in relatively artificial ways (reviewed in Stern & Foster 1996, 1997). In this paper, we look in detail at the aphid Pemphigus spyrothecae, which makes spiral leaf galls on the leaf petioles of Populus nigra L. It has previously been shown in laboratory experiments that the soldiers can kill natural predators and prevent them from entering the galls (Foster 1990). Pemphigus spyrothecae, unlike most Pemphigus species, is non-host-alternating, spending its entire life-cycle on the primary host P. nigra (see Foster & Northcott (1994) for a description of the lifecycle). METHODS We carried out experiments on a row of mature P. nigra var. italica at a site in Cambridge, U.K. (grid ref: TL 485573) on 16–25 September 1993 and 8–21 September 1994. We chose twigs that
? 1998 The Association for the Study of Animal Behaviour
761
762
Animal Behaviour, 55, 3
contained at least three mature, healthy galls of P. spyrothecae. We took a small slice out of the back of each of the three galls and removed all the aphids, then swirled a small camel-hair brush around inside the galls to ensure that no live aphids remained. In the laboratory, we counted samples of aphids collected at the site and placed them in individual tubes: each tube contained either 50 soldiers and 50 non-soldiers or 100 non-soldiers. A full range of non-soldiers was used for each tube, although we avoided fourth instar sexuparae. We did not use any adult (winged) sexuparae in the experiment. The aphids used in the experiments would mostly have been from the third gall generation, although some aphids from the second generation might also have been included (Foster 1990): we made no attempt to distinguish between them. The experimental aphids all came from the same gall, but not from the gall into which they were returned. These tubes were then taken back into the field and, using a minute funnel, we carefully poured the aphids back into the emptied galls. The slice of gall removed earlier was fixed firmly back in place with Blu-Tak, so that the only possible route of entry for predators or aphids was via the gall’s own exit hole(s). On each twig, there was a gall with soldiers and non-soldiers, a gall with nonsoldiers alone and a gall that was enclosed with a close-fitting mesh bag and containing either soldiers and non-soldiers or non-soldiers alone. After 7 days, we collected the galls into individual containers and, in the laboratory, counted the number of live and dead aphids in each gall, taking careful note of whether there was a predator in or around the gall. We also counted the numbers of the aphid Chaitophorus leucomelas, which is commonly found cohabiting in the galls of P. spyrothecae (Foster 1990). We measured the outcomes of these experiments as aphid survival rather than mortality, because counting the number of dead aphids was a relatively poor method of measuring mortality. Many of the aphids eaten by the predators would have left nothing detectable and it was often difficult to distinguish individual dead aphids within the galls. There was no significant difference in the attack frequency in the 2 years (20% (4/20) in 1993 and 16.7% (6/36) in 1994), and there were no significant differences between the 2 years in the
numbers of alive or dead aphids in any of the treatments (for example, comparing 1993 and 1994: with soldiers, total aphids alive, t54 =0.461, ; without soldiers, total aphids alive, t54 =0.30, ). The only significant difference was in the numbers of the cohabiting C. leucomelas, which were much commoner in 1993. All the experimental galls (N=20) were invaded in 1993, but only 11% (4/36) were invaded in 1994. The mean numbers of Chaitophorus in the galls in 1993 & was 20.5&5.1, whereas there were only 0.22&0.17 in 1994 (t54 =5.4, P<0.0001). Since there were no other significant differences in the outcomes of the experiments in the 2 years, the data for the Pemphigus aphids for 1993 and 1994 have been treated together. It is possible that the intrinsic rate of survival of soldiers and non-soldiers in these experiments is different, even in the absence of predators. If this were the case, then it would be invalid to compare galls with and without soldiers: any differences in experimental outcomes might have been caused by different intrinsic rates of survival in soldiers and non-soldiers, rather than by different susceptibilities to predation. It is not possible to compare the survival of non-soldiers and soldiers directly in the current experiment, since there is a probability of up to 40% that a soldier (first instar) will moult into a non-soldier during the week of observation (the first instar lasts approximately 17 days; range 10–32; P. K. Rhoden, personal observations). Observations on the aphids in the bagged galls strongly suggest that the basic survival rates of soldiers and non-soldiers are not significantly different. In the bagged galls, 14 contained a mixture of 50 soldiers and 50 non-soldiers and 14 contained 100 non-soldiers. There was no significant difference in the overall survival in the galls of these two groups of aphids after 1 week: a mean& of 68.7&8.3 of the aphids in the mixed galls and 64.7&7.7 of those in the non-soldier galls survived the experiment (t26 =0.35). This also indicates that there is no significant difference in the propensity of soldiers and non-soldiers to emigrate from the galls. In addition, in the 14 galls that contained 50 soldiers and 50 non-soldiers, there was no significant difference in the numbers of dead soldiers (X&=3.14&1.23) and dead non-soldiers (3.14&0.75) at the end of 1 week (Wilcoxon signed-ranks test: N=14, ). The bagged galls were treated as one unit, since whether they contained just non-soldiers or a
Foster & Rhoden: Soldier aphids and predation Table I. Influence of presence of soldier aphids on likelihood of predator attack in natural field conditions Not attacked by predator
1 9
27 19
Soldiers present No soldiers present
Galls were scored as having been attacked only if a predator was visible in or on the gall at the end of the experiment. Galls contained either 50 soldiers and 50 non-soldiers (soldiers present) or 100 non-soldiers. Likelihood of attack by predators is not independent of the presence of soldiers: Fisher’s exact test: P=0.012.
mixture of soldiers and non-soldiers did not affect the overall survival of the aphid population. Values are given as means&. All tests are unpaired and two-tailed, unless otherwise stated. In all the ANOVA tests used, we carefully checked residuals to ensure that the assumptions of the models (normality and homoscedascity) were not violated.
RESULTS Predation Rate Ten of the 56 experimental galls had predators either in them or on them at the end of the experiment. This suggests that the minimum probability of successful gall attack is about 18% per week. The attack rate on the galls without soldiers (32% per week) is almost certainly a truer estimate of the potential attack rate: the probability of finding predators in the soldier-defended galls at the end of the experiment is likely to be an underestimate of the actual rate of attack, since presumably a number of attacks were successfully repelled by the soldiers in these galls (see Table I). Nature of the Predators The commonest predators were anthocorids (Hemiptera): three adult and one late-instar Anthocoris minki Dohrn; one late-instar Anthocoris nemoralis (Fab.); and two immatures that were too young to identify. Two galls were attacked by syrphid larvae (Diptera): Meliscaeva auricollis (Meigen) and Syrphus ribesii (L.). One gall was attacked by a lacewing larva (Neuroptera) (too young to be identified). The only
80
Number of aphids alive
Attacked by predator
763
60
40
20
0
Bagged
With soldiers
Without soldiers
Figure 1. The mean& number of aphids alive at the end of 1 week in galls of Pemphigus spyrothecae. Each gall originally contained 100 aphids and was either (1) bagged, (2) unbagged, with 50% soldiers, 50% nonsoldiers, or (3) unbagged with 100% non-soldiers. N=28 in each category. The results were analysed as a blocked ANOVA, with the experimental galls in blocks of three on separate branches. There was no effect of branch on live aphid numbers (GLM: F27,54 =0.97, ). Effect of gall type on number of live aphids: blocked ANOVA, GLM: F2,54 =5.04, P<0.01. Non-soldier galls were significantly different from bagged galls (q54 =4.42, P<0.01) and soldier galls (q54 =2.94, P<0.05; Newman–Keuls). Soldier galls and bagged galls were not significantly different (q54 =1.47, ).
predator that successfully invaded a soldiercontaining gall was the syrphid M. auricollis. Soldier Presence and Predator Attack Under natural field conditions, experimental galls that did not contain soldiers were approximately 10 times more likely to show direct evidence of predator attack than were galls that did contain soldiers (32% versus 3.6%; Table I). Soldier presence had a significant effect on the number of live aphids in the galls at the end of the experiment (Fig. 1). There were significantly fewer aphids in the galls without soldiers than in the galls protected by soldiers and those protected by being within a bag. There were more live aphids in the bagged galls than in the soldier-containing galls, but this difference was not significant. The effects shown in Fig. 1 seem to be entirely due to the result of predation in the galls where
Animal Behaviour, 55, 3
764
predators were present at the end of the experiment. If the 18 trios of galls where there was no direct evidence of predation are analysed separately, there is no significant effect of gall type (bagged, with soldiers, without soldiers) on the numbers of live (blocked ANOVA, GLM: F2,34 =1.47, ) or dead (F2,34 =1.30, ) aphids at the end of the experiment. In the 10 galls in which there was direct evidence of predator attack, aphid survival was very low. In four of the galls (including the soldercontaining gall), none of the aphids were left alive; in two of them, there were fewer than 10 aphids left alive; and in all of the galls there were fewer aphids left alive in the attacked galls than in the unattacked, unbagged gall on the same branch. Soldier Presence and Invasion by Chaitophorus Aphids There is no evidence that soldiers are effective in defending the galls against Chaitophorus. Using the 1993 data, the mean number of Chaitophorus aphids was 19.5&5.3 in the soldier-containing galls and 20.5&5.1 in the galls without soldiers (paired t-test: t9 =0.15, ). DISCUSSION Our results clearly establish that, with natural levels of predation, soldiers can be effective in defending aphid galls from predator attack. Earlier laboratory experiments have shown that soldiers are able to prevent specialist gall predators, A. minki, from entering galls of P. spyrothecae (Foster 1990), and it is now clear that this can happen in the field. Predation seems to be an all-or-nothing event: predators either enter the gall and kill a substantial number of aphids or do not enter at all. The role of the soldiers is to stop predators entering the gall. Since the soldiers have a cleaning as well as a defensive role (Benton & Foster 1992), it could be argued that the absence of soldiers might have impaired aphid survival because of increased honeydew accumulation rather than because of increased predation. However, the numbers of aphids in the experimental galls were small, the galls were cleaned out at the start of the experiment, and very little honeydew had accumulated in them after 1 week. In the bagged galls with and
without soldiers, there was no significant difference in aphid survival (see Methods), which strongly suggests that the differences in aphid survival recorded in the experimental galls were due to predation, not effects of different levels of housekeeping. It may be supposed that gall-living aphids would be protected to some extent from predators simply because they are in a gall (Forrest 1987). This may be true and would be worth testing by changing experimentally the amount of protection provided by the gall in a non-soldier-producing species. However, the present observations suggest that the attack rate on gall-living aphids can be very high indeed. On the soldier-free galls, the rate of successful attacks was 28% in 1 week, and one presumes that the actual attack rate might have been much higher. It is possible that the success rate has been inflated somewhat by the experimental techniques used. Fifty soldiers is rather lower than average for galls at this time of year (26 September 1994, mean soldiers per gall=109&21; range 1–248, N=10; P. K. Rhoden personal observations), and the experimental treatment might have made the natural openings to the galls easier to penetrate. However, the results are consistent with Foster’s (1990) observations on P. spyrothecae that on all six sampling occasions between June and August 1989, at least one of the 30 sampled galls contained a live anthocorid, and that on one occasion 20% of the galls had been attacked by A. minki. The high levels of predation are also supported by Dunn’s (1960) observation that in some years within 14 days of gall opening, 90–100% of galls of Pemphigus bursarius had been successfully attacked by anthocorids. This species also has soldiers, although they are less aggressive than those of P. spyrothecae (personal observations). It is therefore clear that we should think of these gall aphids, like aphids in more open situations, as being under constant attack by predators. The predators attacking these galls, given that the sample was relatively small, are surprisingly diverse. Five species were involved: one is a wellknown gall-specialist A. minki (Dusek & Kristek 1959; Lampel 1960; Foster 1990), the others (certainly S. ribesii and A. nemoralis) are probably more generalist predators. The soldiers would seem to be effective against both kinds of predator, but it would be interesting to observe the soldiers’ defensive behaviour in more detail and
Foster & Rhoden: Soldier aphids and predation obtain much more extensive data from natural situations about the effectiveness of the soldiers in relation to different types of predator. This kind of information would be invaluable for the soldier-producing aphids in general and could most readily be obtained from observations on colonies of horned aphids on the secondary hosts. The soldiers do not appear to defend the galls against the cohabiting aphid C. leucomelas. This is surprising, since there is no doubt that this aphid, by producing large quantities of honeydew, has the potential to cause considerable damage to the aphid colonies (Benton & Foster 1992). However, soldiers do not attack Chaitophorus in experimental situations (Foster 1990), and it would seem that they are simply unable to recognize these aphids as a potential threat. It would be worthwhile quantifying the extent of the damage that can be inflicted by Chaitophorus to establish the selective pressure on the soldiers to deal with these aphids.
ACKNOWLEDGMENTS We thank Tim Benton, David Stern and John Whitfield for helpful comments on the manuscript. P.K.R. was supported by a studentship from SERC.
REFERENCES Alexander, R. D., Noonan, K. M. & Crespi, B. J. 1991. The evolution of eusociality. In: The Biology of the Naked Mole-Rat (Ed. by P. W. Sherman, J. U. M.
765
Jarvis & R. D. Alexander), pp. 3–44. Princeton: Princeton University Press. Aoki, S. 1977. Colophina clematis (Homoptera, Pemphigidae), an aphid species with soldiers. Kontyuˆ, 45, 333–337. Benton, T. G. & Foster, W. A. 1992. Altruistic housekeeping in a social aphid. Proc. R. Soc. Lond. Ser. B, 247, 199–202. Bourke, A. F. G. 1997. Sociality and kin selection in insects. In: Behavioural Ecology (Ed. by J. R. Krebs & N. B. Davies), pp. 203–227. Oxford: Blackwells. Choe, J. & Crespi, B. J. 1997. The Evolution of Social Behaviour in Insects and Arachnids. Cambridge: Cambridge University Press. Dunn, J. A. 1960. The natural enemies of the lettuce root aphid Pemphigus bursarius (L.). Bull. entomol. Res., 51, 271–278. Dusek, J. & Kristek, J. 1959. Pozna´mky k vyskytu a bionomii larev pestrenek (Diptera, Syrphidae) v ha´lka´ch topolovych dutilek. Zool. Listy, 22, 299–308. Forrest, J. M. S. 1987. Galling aphids. In: Aphids: their Biology, Natural Enemies and Control, Vol. 2A (Ed. by A. K. Minks & P. Harrewijn), pp. 341–353. Amsterdam: Elsevier. Foster, W. A. 1990. Experimental evidence for effective and altruistic colony defence against natural predators by soldiers of the gall-forming aphid Pemphigus spyrothecae (Hemiptera: Pemphigidae). Behav. Ecol. Sociobiol., 27, 421–430. Foster, W. A. & Northcott, P. A. 1994. Galls and the evolution of social behaviour in aphids. In: Plant Galls (Ed. by M. A. J. Williams), pp. 161–182. Oxford: Clarendon Press. Lampel, G. 1960. Die morphologischen und o¨kologische Grundlagen des Generationswechsels mono¨zicher und hetero¨zischer Pemphiginen der Schwarz- und Pyramidenpappel. Z. angew. Entomol., 47, 334–375. Stern, D. L. & Foster, W. A. 1996. The evolution of soldiers in aphids. Biol. Rev., 71, 27–79. Stern, D. L. & Foster, W. A. 1997. The evolution of sociality in aphids: a clone’s-eye-view. In: The Evolution of Social Behaviour in Insects and Arachnids (Ed. by J. Choe & B. J. Crespi), pp. 150–165. Cambridge: Cambridge University Press.
Soldiers effectively defend aphid colonies against predators in the field WILLIAM A. FOSTER & PHILIP K. RHODEN Department of Zoology, University of Cambridge (Received 22 April 1997; initial acceptance 20 June 1997; final acceptance 28 August 1997; MS. number: 5530)
Abstract. Morphologically specialized soldiers occur in more than 50 aphid species in the families Pemphigidae and Hormaphididae. To study the effectiveness of soldiers of the gall-forming aphid, Pemphigus spyrothecae Pass., in protecting their galls against natural levels of predation, we manipulated the proportions of soldiers and non-soldiers in sets of galls still attached to poplar trees in the field. Galls with 50 soldiers and 50 non-soldiers were approximately 10 times less likely to be attacked by predators than galls that contained 100 non-soldier aphids. There were significantly fewer live aphids, and significantly more dead aphids, in galls without soldiers than in galls protected either by soldiers or by being within a bag. There were no significant differences in the survival of aphids in galls protected by soldiers compared with those protected by bagging. The soldiers did not protect the galls against invasion by the cohabiting aphid Chaitophorus leucomelas Koch. These observations provide the first demonstration that soldiers are effective in defence against natural levels of predation under field ? 1998 The Association for the Study of Animal Behaviour conditions. Communal defence against predators and parasites is one of the most commonly invoked factors favouring the evolution of cooperative behaviour in animals (e.g. Alexander et al. 1991; Bourke 1997). However, morphologically specialized warrior castes are rare in social animals and are found only in termites, several genera of ants and two families of aphids (Choe & Crespi 1997). There are very few experimental studies of the effectiveness of soldiers in any of these animals, although it is generally accepted that the highly specialized soldiers in ants and termites can effectively protect their colonies. Because aphids are generally perceived as being weak and defenceless, and also because the soldiers (unlike those of ants and termites) are first instars and therefore the smallest possible life-history stage, it has been difficult to convince people that they are actually capable of effectively defending their colonies. Aoki (1977), in the original description of a soldier aphid, Colophina clematis, did show that soldiers could kill predators and there have been further laboratory demonstrations since then (summarized by Stern & Foster 1996). Correspondence: W. A. Foster, Department of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, U.K. (email: [email protected]). 0003–3472/98/030761+05 $25.00/0/ar970664
Our aim in this paper is to demonstrate experimentally that aphid soldiers can effectively protect their colonies against predation under field conditions. It is important to establish this point, since all previous experimental studies of predation of soldier-producing aphids have either used predators that are not natural predators of the particular species or have presented natural predators to the galls in relatively artificial ways (reviewed in Stern & Foster 1996, 1997). In this paper, we look in detail at the aphid Pemphigus spyrothecae, which makes spiral leaf galls on the leaf petioles of Populus nigra L. It has previously been shown in laboratory experiments that the soldiers can kill natural predators and prevent them from entering the galls (Foster 1990). Pemphigus spyrothecae, unlike most Pemphigus species, is non-host-alternating, spending its entire life-cycle on the primary host P. nigra (see Foster & Northcott (1994) for a description of the lifecycle). METHODS We carried out experiments on a row of mature P. nigra var. italica at a site in Cambridge, U.K. (grid ref: TL 485573) on 16–25 September 1993 and 8–21 September 1994. We chose twigs that
? 1998 The Association for the Study of Animal Behaviour
761
762
Animal Behaviour, 55, 3
contained at least three mature, healthy galls of P. spyrothecae. We took a small slice out of the back of each of the three galls and removed all the aphids, then swirled a small camel-hair brush around inside the galls to ensure that no live aphids remained. In the laboratory, we counted samples of aphids collected at the site and placed them in individual tubes: each tube contained either 50 soldiers and 50 non-soldiers or 100 non-soldiers. A full range of non-soldiers was used for each tube, although we avoided fourth instar sexuparae. We did not use any adult (winged) sexuparae in the experiment. The aphids used in the experiments would mostly have been from the third gall generation, although some aphids from the second generation might also have been included (Foster 1990): we made no attempt to distinguish between them. The experimental aphids all came from the same gall, but not from the gall into which they were returned. These tubes were then taken back into the field and, using a minute funnel, we carefully poured the aphids back into the emptied galls. The slice of gall removed earlier was fixed firmly back in place with Blu-Tak, so that the only possible route of entry for predators or aphids was via the gall’s own exit hole(s). On each twig, there was a gall with soldiers and non-soldiers, a gall with nonsoldiers alone and a gall that was enclosed with a close-fitting mesh bag and containing either soldiers and non-soldiers or non-soldiers alone. After 7 days, we collected the galls into individual containers and, in the laboratory, counted the number of live and dead aphids in each gall, taking careful note of whether there was a predator in or around the gall. We also counted the numbers of the aphid Chaitophorus leucomelas, which is commonly found cohabiting in the galls of P. spyrothecae (Foster 1990). We measured the outcomes of these experiments as aphid survival rather than mortality, because counting the number of dead aphids was a relatively poor method of measuring mortality. Many of the aphids eaten by the predators would have left nothing detectable and it was often difficult to distinguish individual dead aphids within the galls. There was no significant difference in the attack frequency in the 2 years (20% (4/20) in 1993 and 16.7% (6/36) in 1994), and there were no significant differences between the 2 years in the
numbers of alive or dead aphids in any of the treatments (for example, comparing 1993 and 1994: with soldiers, total aphids alive, t54 =0.461, ; without soldiers, total aphids alive, t54 =0.30, ). The only significant difference was in the numbers of the cohabiting C. leucomelas, which were much commoner in 1993. All the experimental galls (N=20) were invaded in 1993, but only 11% (4/36) were invaded in 1994. The mean numbers of Chaitophorus in the galls in 1993 & was 20.5&5.1, whereas there were only 0.22&0.17 in 1994 (t54 =5.4, P<0.0001). Since there were no other significant differences in the outcomes of the experiments in the 2 years, the data for the Pemphigus aphids for 1993 and 1994 have been treated together. It is possible that the intrinsic rate of survival of soldiers and non-soldiers in these experiments is different, even in the absence of predators. If this were the case, then it would be invalid to compare galls with and without soldiers: any differences in experimental outcomes might have been caused by different intrinsic rates of survival in soldiers and non-soldiers, rather than by different susceptibilities to predation. It is not possible to compare the survival of non-soldiers and soldiers directly in the current experiment, since there is a probability of up to 40% that a soldier (first instar) will moult into a non-soldier during the week of observation (the first instar lasts approximately 17 days; range 10–32; P. K. Rhoden, personal observations). Observations on the aphids in the bagged galls strongly suggest that the basic survival rates of soldiers and non-soldiers are not significantly different. In the bagged galls, 14 contained a mixture of 50 soldiers and 50 non-soldiers and 14 contained 100 non-soldiers. There was no significant difference in the overall survival in the galls of these two groups of aphids after 1 week: a mean& of 68.7&8.3 of the aphids in the mixed galls and 64.7&7.7 of those in the non-soldier galls survived the experiment (t26 =0.35). This also indicates that there is no significant difference in the propensity of soldiers and non-soldiers to emigrate from the galls. In addition, in the 14 galls that contained 50 soldiers and 50 non-soldiers, there was no significant difference in the numbers of dead soldiers (X&=3.14&1.23) and dead non-soldiers (3.14&0.75) at the end of 1 week (Wilcoxon signed-ranks test: N=14, ). The bagged galls were treated as one unit, since whether they contained just non-soldiers or a
Foster & Rhoden: Soldier aphids and predation Table I. Influence of presence of soldier aphids on likelihood of predator attack in natural field conditions Not attacked by predator
1 9
27 19
Soldiers present No soldiers present
Galls were scored as having been attacked only if a predator was visible in or on the gall at the end of the experiment. Galls contained either 50 soldiers and 50 non-soldiers (soldiers present) or 100 non-soldiers. Likelihood of attack by predators is not independent of the presence of soldiers: Fisher’s exact test: P=0.012.
mixture of soldiers and non-soldiers did not affect the overall survival of the aphid population. Values are given as means&. All tests are unpaired and two-tailed, unless otherwise stated. In all the ANOVA tests used, we carefully checked residuals to ensure that the assumptions of the models (normality and homoscedascity) were not violated.
RESULTS Predation Rate Ten of the 56 experimental galls had predators either in them or on them at the end of the experiment. This suggests that the minimum probability of successful gall attack is about 18% per week. The attack rate on the galls without soldiers (32% per week) is almost certainly a truer estimate of the potential attack rate: the probability of finding predators in the soldier-defended galls at the end of the experiment is likely to be an underestimate of the actual rate of attack, since presumably a number of attacks were successfully repelled by the soldiers in these galls (see Table I). Nature of the Predators The commonest predators were anthocorids (Hemiptera): three adult and one late-instar Anthocoris minki Dohrn; one late-instar Anthocoris nemoralis (Fab.); and two immatures that were too young to identify. Two galls were attacked by syrphid larvae (Diptera): Meliscaeva auricollis (Meigen) and Syrphus ribesii (L.). One gall was attacked by a lacewing larva (Neuroptera) (too young to be identified). The only
80
Number of aphids alive
Attacked by predator
763
60
40
20
0
Bagged
With soldiers
Without soldiers
Figure 1. The mean& number of aphids alive at the end of 1 week in galls of Pemphigus spyrothecae. Each gall originally contained 100 aphids and was either (1) bagged, (2) unbagged, with 50% soldiers, 50% nonsoldiers, or (3) unbagged with 100% non-soldiers. N=28 in each category. The results were analysed as a blocked ANOVA, with the experimental galls in blocks of three on separate branches. There was no effect of branch on live aphid numbers (GLM: F27,54 =0.97, ). Effect of gall type on number of live aphids: blocked ANOVA, GLM: F2,54 =5.04, P<0.01. Non-soldier galls were significantly different from bagged galls (q54 =4.42, P<0.01) and soldier galls (q54 =2.94, P<0.05; Newman–Keuls). Soldier galls and bagged galls were not significantly different (q54 =1.47, ).
predator that successfully invaded a soldiercontaining gall was the syrphid M. auricollis. Soldier Presence and Predator Attack Under natural field conditions, experimental galls that did not contain soldiers were approximately 10 times more likely to show direct evidence of predator attack than were galls that did contain soldiers (32% versus 3.6%; Table I). Soldier presence had a significant effect on the number of live aphids in the galls at the end of the experiment (Fig. 1). There were significantly fewer aphids in the galls without soldiers than in the galls protected by soldiers and those protected by being within a bag. There were more live aphids in the bagged galls than in the soldier-containing galls, but this difference was not significant. The effects shown in Fig. 1 seem to be entirely due to the result of predation in the galls where
Animal Behaviour, 55, 3
764
predators were present at the end of the experiment. If the 18 trios of galls where there was no direct evidence of predation are analysed separately, there is no significant effect of gall type (bagged, with soldiers, without soldiers) on the numbers of live (blocked ANOVA, GLM: F2,34 =1.47, ) or dead (F2,34 =1.30, ) aphids at the end of the experiment. In the 10 galls in which there was direct evidence of predator attack, aphid survival was very low. In four of the galls (including the soldercontaining gall), none of the aphids were left alive; in two of them, there were fewer than 10 aphids left alive; and in all of the galls there were fewer aphids left alive in the attacked galls than in the unattacked, unbagged gall on the same branch. Soldier Presence and Invasion by Chaitophorus Aphids There is no evidence that soldiers are effective in defending the galls against Chaitophorus. Using the 1993 data, the mean number of Chaitophorus aphids was 19.5&5.3 in the soldier-containing galls and 20.5&5.1 in the galls without soldiers (paired t-test: t9 =0.15, ). DISCUSSION Our results clearly establish that, with natural levels of predation, soldiers can be effective in defending aphid galls from predator attack. Earlier laboratory experiments have shown that soldiers are able to prevent specialist gall predators, A. minki, from entering galls of P. spyrothecae (Foster 1990), and it is now clear that this can happen in the field. Predation seems to be an all-or-nothing event: predators either enter the gall and kill a substantial number of aphids or do not enter at all. The role of the soldiers is to stop predators entering the gall. Since the soldiers have a cleaning as well as a defensive role (Benton & Foster 1992), it could be argued that the absence of soldiers might have impaired aphid survival because of increased honeydew accumulation rather than because of increased predation. However, the numbers of aphids in the experimental galls were small, the galls were cleaned out at the start of the experiment, and very little honeydew had accumulated in them after 1 week. In the bagged galls with and
without soldiers, there was no significant difference in aphid survival (see Methods), which strongly suggests that the differences in aphid survival recorded in the experimental galls were due to predation, not effects of different levels of housekeeping. It may be supposed that gall-living aphids would be protected to some extent from predators simply because they are in a gall (Forrest 1987). This may be true and would be worth testing by changing experimentally the amount of protection provided by the gall in a non-soldier-producing species. However, the present observations suggest that the attack rate on gall-living aphids can be very high indeed. On the soldier-free galls, the rate of successful attacks was 28% in 1 week, and one presumes that the actual attack rate might have been much higher. It is possible that the success rate has been inflated somewhat by the experimental techniques used. Fifty soldiers is rather lower than average for galls at this time of year (26 September 1994, mean soldiers per gall=109&21; range 1–248, N=10; P. K. Rhoden personal observations), and the experimental treatment might have made the natural openings to the galls easier to penetrate. However, the results are consistent with Foster’s (1990) observations on P. spyrothecae that on all six sampling occasions between June and August 1989, at least one of the 30 sampled galls contained a live anthocorid, and that on one occasion 20% of the galls had been attacked by A. minki. The high levels of predation are also supported by Dunn’s (1960) observation that in some years within 14 days of gall opening, 90–100% of galls of Pemphigus bursarius had been successfully attacked by anthocorids. This species also has soldiers, although they are less aggressive than those of P. spyrothecae (personal observations). It is therefore clear that we should think of these gall aphids, like aphids in more open situations, as being under constant attack by predators. The predators attacking these galls, given that the sample was relatively small, are surprisingly diverse. Five species were involved: one is a wellknown gall-specialist A. minki (Dusek & Kristek 1959; Lampel 1960; Foster 1990), the others (certainly S. ribesii and A. nemoralis) are probably more generalist predators. The soldiers would seem to be effective against both kinds of predator, but it would be interesting to observe the soldiers’ defensive behaviour in more detail and
Foster & Rhoden: Soldier aphids and predation obtain much more extensive data from natural situations about the effectiveness of the soldiers in relation to different types of predator. This kind of information would be invaluable for the soldier-producing aphids in general and could most readily be obtained from observations on colonies of horned aphids on the secondary hosts. The soldiers do not appear to defend the galls against the cohabiting aphid C. leucomelas. This is surprising, since there is no doubt that this aphid, by producing large quantities of honeydew, has the potential to cause considerable damage to the aphid colonies (Benton & Foster 1992). However, soldiers do not attack Chaitophorus in experimental situations (Foster 1990), and it would seem that they are simply unable to recognize these aphids as a potential threat. It would be worthwhile quantifying the extent of the damage that can be inflicted by Chaitophorus to establish the selective pressure on the soldiers to deal with these aphids.
ACKNOWLEDGMENTS We thank Tim Benton, David Stern and John Whitfield for helpful comments on the manuscript. P.K.R. was supported by a studentship from SERC.
REFERENCES Alexander, R. D., Noonan, K. M. & Crespi, B. J. 1991. The evolution of eusociality. In: The Biology of the Naked Mole-Rat (Ed. by P. W. Sherman, J. U. M.
765
Jarvis & R. D. Alexander), pp. 3–44. Princeton: Princeton University Press. Aoki, S. 1977. Colophina clematis (Homoptera, Pemphigidae), an aphid species with soldiers. Kontyuˆ, 45, 333–337. Benton, T. G. & Foster, W. A. 1992. Altruistic housekeeping in a social aphid. Proc. R. Soc. Lond. Ser. B, 247, 199–202. Bourke, A. F. G. 1997. Sociality and kin selection in insects. In: Behavioural Ecology (Ed. by J. R. Krebs & N. B. Davies), pp. 203–227. Oxford: Blackwells. Choe, J. & Crespi, B. J. 1997. The Evolution of Social Behaviour in Insects and Arachnids. Cambridge: Cambridge University Press. Dunn, J. A. 1960. The natural enemies of the lettuce root aphid Pemphigus bursarius (L.). Bull. entomol. Res., 51, 271–278. Dusek, J. & Kristek, J. 1959. Pozna´mky k vyskytu a bionomii larev pestrenek (Diptera, Syrphidae) v ha´lka´ch topolovych dutilek. Zool. Listy, 22, 299–308. Forrest, J. M. S. 1987. Galling aphids. In: Aphids: their Biology, Natural Enemies and Control, Vol. 2A (Ed. by A. K. Minks & P. Harrewijn), pp. 341–353. Amsterdam: Elsevier. Foster, W. A. 1990. Experimental evidence for effective and altruistic colony defence against natural predators by soldiers of the gall-forming aphid Pemphigus spyrothecae (Hemiptera: Pemphigidae). Behav. Ecol. Sociobiol., 27, 421–430. Foster, W. A. & Northcott, P. A. 1994. Galls and the evolution of social behaviour in aphids. In: Plant Galls (Ed. by M. A. J. Williams), pp. 161–182. Oxford: Clarendon Press. Lampel, G. 1960. Die morphologischen und o¨kologische Grundlagen des Generationswechsels mono¨zicher und hetero¨zischer Pemphiginen der Schwarz- und Pyramidenpappel. Z. angew. Entomol., 47, 334–375. Stern, D. L. & Foster, W. A. 1996. The evolution of soldiers in aphids. Biol. Rev., 71, 27–79. Stern, D. L. & Foster, W. A. 1997. The evolution of sociality in aphids: a clone’s-eye-view. In: The Evolution of Social Behaviour in Insects and Arachnids (Ed. by J. Choe & B. J. Crespi), pp. 150–165. Cambridge: Cambridge University Press.