BehavioralEcology Behav Ecol Sociobiol (1987) 21:401-406 and Sociobiology ? Springer-Verlag1987

Host location and exploitation by the cleptoparasiticwasp Argochrysis armilla: the role of learning(: Chrysididae)

Jay A. Rosenheim Departmentof Entomological Sciences, University of California, Berkeley,CA 94720, USA

Received May 27, 1987 / Accepted August 31, 1987

Summary. The nesting behaviors of many solitary (Evans 1977); but it is also significant in temporal- ground-nesting incorporate temporal bar- ly separating the stage of the nesting cycle that riers against would-be cleptoparasites. Nests being is most conspicuous to parasites, nest digging, excavated are conspicuous but relatively invulner- from the stage that is most vulnerable to exploita- able to parasites, while nests being provisioned, tion, nest provisioning (Rosenheim 1987a). In this often several hours to days later, are inconspicuous way host behavior challenged potential cleptopara- but highly vulnerable. Argochrysisarmilla, a clep- sites to bridge the temporal gap between nest exca- toparasite of solitary ground-nesting wasps, Am- vation and nest provisioning. Here I present exper- mophila spp., bridges the temporal gap between imental results for one cleptoparasite, Argochrysis nest excavation and provisioning by (i) visually lo- armilla Bohart (Hymenoptera: Chrysididae), that cating digging hosts, (ii) learning the locations of has adapted to this temporal gap; Argochrysisar- associated nests, (iii) maintaining surveillance on milla (i) orients visually to digging hosts, (ii) learns a series of nests during the hosts' absence, and the locations of associated nests, (iii) attends a se- (iv) ovipositing in nests when the host returnswith ries (or "trapline") of nests for up to several days provisions. Patterns of surveillanceand parasitism while the host hunts for provisions, and (iv) suc- of Ammophiladysmica nests were generated by the cessfully oviposits in attended nests during nest number of cleptoparasitesdiscovering and learning provisioning. Observations of a nesting aggrega- the nest's location during excavation. These results tion of Ammophiladysmica Menke (Hymenoptera: support recent suggestions that learning may play ) revealed the significance of the learned an important role in shaping foraging strategies foraging strategy in generatingpatterns of nest sur- of parasites. veillance and parasitism. This study is the first demonstrating a strategy of host exploitation based upon locality learning by an insect parasite and supports recent suggestions of the importance of in insect and Rausher Introduction learning foraging (Papaj 1983; Gould 1985, 1986; Menzel 1985; Lewis Early comparative ethologists used landmark dis- 1986; Papaj 1986; Jermy 1986). placement techniques to establish the ability of sol- the of itary digger wasps to learn "topography" Methods their nest-site and thereby relocate previously con- structed nests (van Iersel 1975). Locality learning Argochrysisarmilla was studied from 1982-1986 at Sagehen Creek Field Nevada where enabled to change Station, County, California, USA, solitary ground-nesting wasps it developed as a cleptoparasitein the nests of solitary ground- their nesting behavior from the primitive prey-nest nesting wasps of the genus Ammophila.The study site was lo- sequence, in which the nest is excavated after the cated on a broad ridgetop, elevation 2000 m, where Ammophila provisions are collected, to the more advanced spp. nested in several aggregations along a dirt road. Ammo- nests with one to several nest-prey sequence, in which the frequentlylengthy phila provision unicellular lepidop- after the nest is teran larvae, and both species discussed here, Ammophiladys- hunting process begins dug (Evans mica and Ammophilaazteca Cameron, exhibited nest-prey be- 1958). This reordering avoids the risk of theft or havior patterns.Argochrysis armilla achieved annual parasitism parasitism of the provisions during nest excavation rates of 21.3-43.5% in nests of Ammophiladysmica, despite

This content downloaded from 128.120.194.195 on Tue, 23 Sep 2014 01:53:03 AM All use subject to JSTOR Terms and Conditions 402 the fact that nest provisioning occurred an average of host's absence. Observationsmade before this experimenthad 10.4 + 8.5 h after nest excavation (n = 128) (Rosenheim 1987a). indicated that only parasites that had discovered a nest during No other parasites achieved parasitism rates in excess of 4%. excavationwere able to reorientto the nest in the host's absence For a more detailed descriptionof the site, the nesting behavior (see below). Ammophilaazteca initiating nest excavations were of Ammophiladysmica, and interactionsof Ammophiladysmica surroundedby four artificiallandmarks (white plastic LEGO? with Argochrysisarmilla see Rosenheim (1987a). blocks, 32 x 16 x 19 mm with eight small cylindrical knobs on the upper surface) each placed 3 cm from the nest entrance in a Parasite surveillance host nests square array. These landmarksremained in place for the of duration of the digging. The presence and identity of marked The abundance of Argochrysisarmilla attending Ammophila Argochrysisarmilla as well as the presence of unmarkedpara- dysmica nests was assessed during three stages of the nesting sites attracted to the nest were recorded. Parasite identity was cycle by scoring the maximum number of parasites present si- recorded as a means of (i) distinguishing between initial and multaneouslyat any time during the samplingperiod. Parasites subsequentvisits and (ii) increasingthe replicatenumber, since were consideredto be in attendenceif they landed on or hovered only the first return of an unmarkedparasite could be scored directly above the nest entrance, or if they perched at least as an initial return, and all subsequentvisits by any unmarked twice sequentially facing the nest entrance from a distance of parasites had to be lumped in the total visits column. At the less than 30 cm. The period of nest digging was monitored conclusion of the nest excavationthe host and parasitesgeneral- to yield the first sample; nest digging required an average of ly departed. If all Argochrysisarmilla had not departed within 63.5 + 35.6 minutes (n = 102). During the absence of the hunting ten minutes, those remainingwere flushed from the nest. Two , 3-min nest surveys were performed hourly during the false nests, each consisting of a short vertical tunnel provided period of high parasite activity, 1100-1600 h. These surveys with a single loose-fitting pebble closure similar to that of Am- were made for five days after nest excavation or one day after mophilaazteca, were then constructed6 cm from, and on oppo- the completion of nest provisioning, whichever occurred first. site sides of, the true nest. The artificial landmarkswere then Finally, parasite abundance was monitored during the period removed. A coin flip determinedwhich of the two false nests of the nest's greatest vulnerability,when the nest closure had would be surrounded by an identical array of clean artificial been removed by the host wasp to deposit the caterpillarpro- landmarks. During the following hour, or until the return of vision, oviposit, clean the nest, and find a pebble with which the host, returnsof Argochrysisarmilla to membersof the nest to re-plug the nest. The duration of this period averaged array were scored when parasites hovered directly over or 185.7+ 130.3 s (n= 50). Because Argochrysisarmilla quit the landed on a nest. The scoring method was chosen because it nest-site after ovipositing in a host nest, the parasiteabundance was unambiguous and avoided the possibility of false positive score during provisioning was slightly modified: if the maxi- scores associated with attraction to the LEGO? blocks inde- mum number of parasitessimultaneously present was observed pendent of their role as landmarks. Twelve trials were con- after n parasites had oviposited and departed, the score was ducted from 1 July 1986 to 14 July 1986. increasedby n. At the end of the nesting season all Ammophiladysmica nests were excavated and scored as parasitized (Argochrysis Results armilla cocoon(s) present) or unparasitized(host cocoon pres- ent); nests whose contents were destroyed by mold or other Host Location parasitesor predatorswere not included in the analysis. A sample of Argochrysisarmilla was marked individually The manipulated Ammophilalure employed in the with enamel paints spotted onto the thoracic dorsum (n =47). attraction experiment increased both the number The identity of these individuals could be determinedwithout disturbancein the field. of Argochrysis armilla and the duration of their stay in the experimental plots (Table 1). Parasites Host location were observed to reposition themselves on their perches to continue facing the moving lure. Ar- Digging Ammophiladysmica were rapidly discovered by Ar- armilla within the gochrysis armilla, most nests being attended by one or more gochrysis searching nesting ag- parasiteswithin the first 10 min of digging (Rosenheim 1987b). gregation detected the moving lure only if they To determine if parasites were using visual cues to orient to were perched nearby (maximum distance of detec- digging wasps, an artificial Ammophilalure was employed in tion appeared to be ca. 0.5 m) and were facing an attraction experiment.The lure consisted of a dead female towards the lure. Parasites in approximately moving Ammophilaazteca sealed a coat of clear nail polish and sus- the lure in low were not ar- pended from a pole by a clear line. Fifty circular plots, each past cruising flights 0.5 m in diameter, were established along a 25 m transect of rested, a result confirmed by observations of dig- a nesting aggregation and randomly assigned to either the ex- ging Ammophila.Although this experimenthas not perimental (lure manipulated to simulate digging movements) ruled out complementary roles for chemical or or control treatment Treatments were for (no lure). applied other cues, visual cues to be sufficient ten minutes during which parasite abundance (scored as the appeared maximumnumber of parasites simultaneouslypresent) and the to mediate initial host location by Argochrysisar- duration of the presence of > 1 parasite within the plot was milla. recorded. These results alone did not, however, explain the observed rates of nest parasitism. Nests under Locality learning construction were not generally vulnerable to Ar- A landmarkdisplacement experiment was performedto investi- gochrysis armilla; parasites successfully penetrated gate the mechanism of parasite orientation to nests during the the nest (a prerequisiteof oviposition) during only

This content downloaded from 128.120.194.195 on Tue, 23 Sep 2014 01:53:03 AM All use subject to JSTOR Terms and Conditions 403

Table 1. Argochrysisarmilla abundance and time present in plots (s.) with and without a manipulatedAmmophila lure

Treatment Parasiteabundance score time (mean present per plot) Nest 0 1 2

No lure 19 6 0 Departing (-) (8.9+ 5.0s) (- parasite Lure 6 10 9 Perch (-) (103.6+63.4 s) (246.8+131.3 s)

Parasiteabundance and time present, lure vs. no lure, ts= 4.01, Fig. 1. Argochrysisarmilla orientation flight. Circlingflight was P<0.001; and ts=4.16, P<0.001, respectively, Wilcoxon 2- rapid and close to the ground (<5 cm). Variable aspects in- sample test cluded the diameter of the circles (usual range ca. 10-100 cm, rarely up to 180 cm) and the number of circles flown (usual range ca. 1-5, rarely up to 12)

3 of the 109 Ammophiladysmica nest excavations Table 2. Landmarkdisplacement experiment. Presented are in- observed in 1986. Eggs deposited in nests being dividual parasite returns to members of an experimentalarray of nests the true a false nest surrounded would likely have been quickly discarded with including: (i) nest, (ii) dug by landmarks displaced from their former positions around excavated dirt. The parasites' strategy of host ex- the true nest, and (iii) a false nest without landmarks ploitation therefore had to extend beyond the ini- tial nest discovery. Nest visited Individually marked parasites were commonly (i) (ii) (iii) observed perching around and investigating the True nest False nest False nest closures of nests while the host was away hunting with landmarks without landmarks for provisions. These nest-attendingparasites were present at a nest only intermittently during the First return 4 n.s. 12 *** 0 also returned to nests from their Second return 1 n.s. 7 * 0 day; they nightly Total returns 9 *** 44 *** 1 sleeping aggregations for up to 5 days after the nest excavation. Fifty-eight observations of Exact binomial probabilities were calculated. Following Bon- marked parasites attending nests in the absence ferroni's inequality for 2 pairwise comparisons for each class of the host were made for nests (n = 18) at which of returns, differenceswere considered non-significant(n.s.) if *: ***: P<0.0005 the identity was known of all marked parasites P>0.025; P<0.025; that had discovered the nest during the digging Without stage. exception, nest-attending parasites surrounded the landmarksthan to the true nest. had also been earlier nest by present during digging. Marked not the nest to be a parasites present during dig- Thus, prior experience appeared prerequi- did not score visits. All visits site of reorientation to nests in the host's absence. ging any positive of individual parasites were made to the same nest, In addition, parasites leaving newly discovered suggesting that parasites visiting the false nest were nests for the first time displayed stereotyped cir- not alerted to their mistake by the absence of cues cling flights centered on the nest entrance (Fig. 1). normally associated with a true nest. This con- Finally, it was parasites present at the nest when trasted with the behavior of the host wasps, who the host returned with that produced provisions chose the false nest surrounded by the landmarks the 21-43% rates observed; Argochrysis parasitism in 5 of 6 instances, but in 3 of these 5 cases quickly armilla during 32 of 94 (34.0%) Ammo- oviposited shifted to the true nest after briefly antennating phila dysmica nest provisionings observed in 1986. the false nest's closure. Thus, the parasites ap- peared to return to the nest by using a learned Locality learning knowledge of the nest site's topography, including Parasites returning to the 3-nest array of the land- the location of the nest relative to the artificial mark displacement experiment made a significant- landmarks. ly greater number of first, second, and total returns to the false nest surrounded the land- by displaced Parasite surveillance host nests marks than to the false nest without landmarks of (Table 2). Parasites also made a significantly Field observations of marked Argochrysisarmilla greater number of total returns to the false nest explained the intermittent nature of nest surveil-

This content downloaded from 128.120.194.195 on Tue, 23 Sep 2014 01:53:03 AM All use subject to JSTOR Terms and Conditions 404

1.60

1.28.. \ Fig. 2. Nest surveillanceby Argochrysis armilla. Average parasite abundance scores during the host's absence versus - 0.96 \ plotted time since the completion of nest digging. Time calculated using the 9-h of o day Ammophila dysmica activity. Nests grouped by their ro 06 \ parasite abundance scores during digging: closed circles score of 3-5 (n = 21); triangles score of 2 (n= " 0.32 -, 35); squares score of 1 (n = 27); open circles score of 0 (n = 22). . ~ ._ ...... ,.><;_...,^,, Eachpoint representsan average O 6CI 6_? _ YA 6 of 5-21 observations 1-3 4-6 7-9 10-12 13-15 16-18 19-21 22-24 25-34 35-44 45-54 Time (hours after nest digging)

lance. Individual parasites were observed to inter- 1.22 + 1.39 (n = 9), respectively. Nest outcome was rupt their surveillanceof one nest to fly to another also strongly related to the parasite abundance active nest, where they would perch or inspect the score during digging, with the probability of nest closure for a variable time and then either return parasitism rising from 7% to 60% as parasite to the first nest or fly off to a third. Preliminary abundance during digging rose from 0 to > 3 (Ro- observations revealed parasites travelling as far as senheim 1987b). Thus, it was variation in the 42 m to traverse a trapline of up to 4 nests. number of parasites discovering nests during exca- To relate the strategy of host exploitation ex- vation that generated the observed patterns of nest hibited by Argochrysisarmilla to the ecological im- surveillance during the host's absence, parasite pact of this parasite upon Ammophiladysmica, par- presence during nest provisioning, and nest para- asite abundance during nest excavation was com- sitism. The ability to learn the location of dis- pared to parasite abundance during later stages covered nests and thereby reorient to nests in the of the nesting cycle and to the final nest outcome. host's absence enabled Argochrysis armilla to Patterns of nest surveillanceby Argochrysisarmilla bridge the temporal gap between nest digging and during the host's hunting period are presented in nest provisioning. The incorporation of several Fig. 2. The apparent trends were tested for statisti- nests into a trapline may have increased the effi- cal significancein the following way: (i) each nest's ciency of the nest-attendingprocess. parasite abundance scores during the host's ab- sence were versus time a least (y) plotted (x), (ii) Discussion squares line was fitted to each plot, (iii) the y- intercept and slope of these lines were then tested The landmark displacement experiment distin- for correlation to the nest's correspondingparasite guished between two possible mechanisms of Ar- abundance score during digging. Nests with higher gochrysis armilla reorientation that incorporated parasite abundance scores during digging had the observed role of prior experience with the nest, larger parasite abundance scores during the host's i.e., that (i) parasites deposited a chemical mark absence (Spearman's rank correlation; rs=0.524, on or near nests to which they could subsequently P<0.001), these scores declining more rapidly with orient, or (ii) that parasites learned the nest's loca- time (rs= -0.218, P<0.02). Parasite abundance tion. The preference of parasites for the false nest scores during the critical stage of nest provisioning surrounded by the displaced landmarks indicates were also correlated with parasite abundance dur- that a learned knowledge of the nest's location is ing excavation (r,=0.287, P=0.05). Nests with the primary means of orientation in the immediate parasite abundance scores during digging of 0, 1, vicinity of the nest. The observed returns to the 2, and 3-5 had parasite abundance scores during true nest may be attributableto the abundant natu- provisioning (mean+SD) of 0.57+0.98 (n=7), rally occurring landmarks, such as rocks or grass 0.70+0.95 (n=10), 1.52+2.32 (n=21), and clumps, as well as possible chemical cues which

This content downloaded from 128.120.194.195 on Tue, 23 Sep 2014 01:53:03 AM All use subject to JSTOR Terms and Conditions 405 may have competed with the artificial landmarks sis armilla appears therefore to be a trait derived for the parasite's attention. within the Chrysididae.Hosts excavating nests be- Many elements of the foraging and host utiliza- fore capturing prey and hosts constructing multi- tion strategies of insect parasites may be modified cellular nests may have generated significant selec- by experience. Parasites may learn novel responses tion pressures favoring the locality learning trait. to visual or chemical cues associated with the host Locality learning and/or traplining has been or host's microhabitat (Taylor 1974; Arthur 1981; described in a wide array of insect groups, includ- Vet and van Opzeeland 1984; Wardle and Borden ing pollinators, herbivores, nest-builders, species 1985) and may also learn to discriminate between using territorialor lekking mating systems, species parasitized and unparasitized hosts (van Lenteren sleeping in aggregations (Wehner 1981), and now 1981). Locality learning has not, however, been in a parasite, Argochrysis armilla. The common previously demonstratedfor an insect parasite. Ob- ecological factor underlying the adaptive signifi- servations of foraging by members of several gen- cance of this learning behavior in these groups ap- era of cleptoparasitic do, however, suggest pears to be the need to return to temporally stable that locality learning may be widely employed as and spatially heterogeneous resources (Baker a means of returning to host nests (Linsley and 1978). These are distributional characteristics of MacSwain 1955; Rozen et al. 1978; Eickwort and many insect populations (Hassell 1978), which are Abrams 1980; Cane 1983). Parasitic bees perform the key resources of insect parasites. Locality circling "orientation" flights around host nests, learning may, therefore, be a potentially adaptive travel between multiple nests, and return to pre- trait for many insect parasites. viously located nests. Whether the basis for these behaviors is a learned knowledge of nest location Acknowledgements.I thank T. Meade and I.G. Powch for dedi- is not known. Because bees evolved re- cated assistance in the field. My gratitude is also extended to parasitic M.P. and staff di- from of bees J.P. Thornton, Yoder-Williams, the and cently groups nest-building (Bohart rectorshipof the SagehenCreek Field Station (Univ. of Califor- 1970), their putative locality learning ability prob- nia). R.M. Bohart kindly confirmed the identity of the wasps ably preceded their parasitic habit. Learning there- studied. For encouragementand guidance I thank M.A. Hoy fore may explain the observed patterns of clepto- and P.S. Ward. The manuscript was critically reviewed and improved by E.A. Bernays, L.E. Caltagirone, T. Meade, L.E. parasitic foraging. Newstrom, I.G. Powch, B.H. Smith, M.R. Strand, C.A. Toft, Locality learning for Argochrysis armilla ap- and P.S. Ward. This material is based upon work supported pears in contrast to be a trait derived within the in part by a Grant-in-Aid from the Graduate Division, U.C. Chrysididae. The , comprising the Berkeley,and under an NSF Graduate Fellowship. Plumariidae, , , Em- bolemidae, , , and Chrysidi- References dae, diverged early from the evolutionary lineage of the aculeate a mono- Arthur AP (1981) Host acceptance by . In: Nord- Hymenoptera, forming lund DA, Jones RL, Lewis WJ (eds) Semiochemicals,their phyletic sister group to the remaining role in pest control. Wiley, New York, pp 97-120 (Brothers 1975; Carpenter 1986). Host searching Baker RR (1978) The evolutionary ecology of migra- within a wood or soil substrate appears likely to tion. Holmes & Meier, New York have been the habit for the Aculeata and Bohart GE (1970) The evolution of parasitism among bees. primitive Fac Honor Lect, Utah State University, Logan, Utah the Chrysidoidea (Rasnitsyn 1980; Carpenter BrothersDJ (1975) Phylogeny and classificationof the aculeate 1986). This habit appears to be retained by the Hymenoptera,with special referenceto the .Univ primitive chrysidoid families, the Plumariidae and Kansas Sci Bull 50:483-648 Scolebythidae, and is an unlikely setting for the Cane JH (1983) Olfactoryevaluation of Andrenahost suitability by kleptoparasiticNomada bees (Hymenoptera: ). evolution of locality learning as a foraging strate- Anim Behav 31:138-144 gy, due to the constraints on vision. Only within CarpenterJM (1986) Cladisticsof the Chrysidoidea(Hymenop- the bethylid genus Epyris do we find nest building tera). J New York Entomol Soc 94:303-330 in the Chrysidoidea (Rubink and Evans 1979); in Eickwort GC, Abrams J (1980) Parasitism of sweat bees in this case the nest is excavated after the the genus Agapostemonby cuckoo bees in the genus Nomada capturing (Hymenoptera: , Anthophoridae). Pan-Pacific single prey, and the prey is carried by the nest-site Entomol 56:144-152 searching wasp, eliminating the need for a learned Evans HE (1958) The evolution of social life in wasps. Proc knowledge of the nest's location. Thus locality Intern Congr Entomol 10th, Montreal, 1956, vol 2, pp 613- within either the context of or 617 learning, foraging HE Extrinsicversus intrinsicfactors in the evolu- return to a is unknown within the Evans (1977) nest, Chrysidoi- tion of insect . Bioscience27:613-617 dea exclusive of Argochrysis armilla. Based upon Gould JL (1985) How bees rememberflower shapes. Science parsimony alone, the learning ability of Argochry- 227:1492-1494

This content downloaded from 128.120.194.195 on Tue, 23 Sep 2014 01:53:03 AM All use subject to JSTOR Terms and Conditions 406

Gould JL (1986) The locale map of honey bees: do RasnitsynAP (1980) The origin and evolution of Hymenoptera. have cognitive maps? Science 232:861-863 Trudy Paleontol Inst 174:1-190 [in Russian] Hassell MP (1978) The dynamics of predator-prey Rosenheim JA (1987a) Nesting behavior and bionomics of a systems. Princeton,New Jersey solitary ground-nestingwasp, Ammophiladysmica (Hyme- Iersel JJA van (1975) The extension of the orientation system noptera: Sphecidae):the influenceof parasitepressure. Ann of Bembix rostrata as used in the vicinity of its nest. In: Entomol Soc Am (in press) BaerendsG, Beer C, Manning A (eds) Function and evolu- Rosenheim JA (1987b) Parasite presence acts as a proximate tion in behaviour. Clarendon,Oxford, pp 142-168 cue in the nest-site selection process of the solitary digger Jermy T (1986) The role of experience in the host selection wasp, Ammophiladysmica (Hymenoptera: Sphecidae). J In- of phytophagous insects. In: Chapman RF, Bernays EA, sect Behav (in press) Stoffolano JG (eds) Perspectivesin chemoreceptionand be- Rozen JG Jr, Eickwort KR, Eickwort GC (1978) The bionom- havior. Springer,Berlin HeidelbergNew York, pp 143-157 ics and immature stages of the cleptoparasitic bee genus Lenteren JC van (1981) Host discrimination by parasitoids. Protepeolus(Anthophoridae, Nomadinae). Am Mus Novi- In: Nordlund DA, Jones RL, Lewis WJ (eds) Semiochemi- tates, no. 2640, pp 1-24 cals, their role in pest control. Wiley, New York, pp 153-179 Rubink WL, Evans HE (1979) Notes on the nesting behavior Lewis AC (1986) Memory constraintsand flower choice in Pier- of the bethylid wasp, Epyris eriogoni Kieffer, in Southern is rapae. Science 232:863-865 Texas. Psyche 86:313-319 Linsley EG, MacSwain JW (1955) The habits of Nomada opa- Taylor RJ Q974) Role of learning En insect parasitism. Ecol cella Timberlakewith notes on other species (Hymenoptera: Monogr 44:89-104 Anthophoridae).Wasmann J Biol 13:253-276 Vet LEM, OpzeelandK van (1984) The influence of condition- Menzel R (1985) Learning in honey bees in an ecological and ing on olfactory microhabitatand host location in Asobara behavioralcontext. In: Holldobler B, LindauerM (eds) Ex- tabida(Nees) and A. rufescens(Foerster) (:Aly- perimental behavioral ecology and sociobiology. Sinauer, siinae) larval parasitoidsof Drosophilidae. Oecologia (Ber- Sunderland,pp 55-74 lin) 63:171-177 Papaj DR (1986) Irfterpopulationdifferences in host preference Wardle AR, Borden JH (1985) Age-dependent associative and the evolution of learning in the butterfly, Battus phi- learningby Exeristesroborator (F.) (Hymenoptera:Ichneu- lenor. Evolution 40:518-530 monidae). Can Entomol 117:605-616 Papaj DR, Rausher MD (1983) Individual variation in host Wehner R (1981) Spatial vision in . In: Autrum H location by phytophagousinsects. In: Ahmad S (ed) Herbiv- (ed) Handbook of sensory physiology, vol VII/6C. Inverte- orous insects, host-seeking behavior and mechanisms.Aca- bratevisual centersand behaviorII. Springer,Berlin Heidel- demic Press, New York, pp 77-124 berg New York, pp 287-616

This content downloaded from 128.120.194.195 on Tue, 23 Sep 2014 01:53:03 AM All use subject to JSTOR Terms and Conditions