Much of the interest in dutch size evolution involves parasitic

Transcription

1 Behavioral Ecology Vol. 8. 6: 647-6M Environmental and genetic determinants of ovicide in the parasitic wasp Bracon hebetor Kegi Takasu,* Paul J. Ode,* Michael F. Antofin, b and Michael R. Strand* Department of Entomology, University of Wisconsin-Madison, Madison, WI 53706, USA and "Department of Biology, Colorado State University, Fort Collins, CO 80523, USA Superparasitism occurs when a parasitoid lays a second dutch of eggs on a host previously parasitized by herself or a conspecific. Ovidde refers to a parasitoid destroying an ^H«ring dutch of eggs on a parasitized host before laying a second clutch. We investigated environmental and genetic determinants of ovidde in the parasitic wasp Bracon hebetor. Characterization of egglaying behavior revealed that B. hebetor commits ovidde during the host examination phase of oviposition. The temporal costs of ovidde were found to be relatively small for females that experienced low rates of host encounter, whereas the costs of ovidde increased for females that experienced a high rate of host encounter. Individual wasps committed ovidde on conspecifically parasitized hosts more frequently than on self-parasitized hosts. Manipulation experiments suggested that B. hebetor females learn about their environment while foraging and commit ovidde on die basis of die travel time between successive hosts. Significant differences were also found in oviddal behavior among laboratory and field populations of B. hebetor. The implications of our results for dutch size theory and die evolution of ovidde are discussed. Key words: Bracon hebetor, competition, foraging, infantidde, ovidde, parasitoid, reproduction, superparasitism. [Behav Ecol 8: (97)] Much of the interest in dutch size evolution involves parasitic wasps (Le., parasitoids). Parasitoids lay their eggs in or on other arthropods, and their progeny develop by consuming and killing die host. The amount of resource available to each offspring is a function of both die quality of die host and the number of eggs laid upon it. Thus, which hosts a female chooses to parasitize and die number of eggs she lays per host will strongly influence die fitness of her offspring (Charnov and Skinner, 88; Godfray, 94; Iwasa et al., 84; Mangel eta]., 94). A common feature of parasitoid biology is that females encounter both unparasitized and parasitized hosts while foraging. Laying a second dutch of eggs onto a host parasitized by the same spedes of parasitoid is called superparasitism. A large literature documents that many parasitoids superparasitize hosts under certain circumstances (Godfray, 94; van Alphen and Visser, 90). Because die eggs in a first dutch will usually hatch first and consume a portion of the host, parasitized hosts are usually of lower quality than comparablysized unparasitized hosts. In addition, die progeny in a second dutch are usually at a competitive disadvantage to individuals in the first dutch. Superparasitism is predicted, therefore, to increase under conditions where die availability of unparasitized, higher quality hosts is low, females have large numbers of eggs available for oviposition, and progeny in a second dutch have some probability of survival (Iwasa et al., 84; Mangel, 92; Roitberg et al, 92; Skinner, 85; Strand and Godfray, 89; Visser et al, 92; Weisser and Houston, 93). One strategy for increasing die value of a parasitized host would be for a second female to kill die progeny in a first dutch by biting diem with her mandibles or piercing diem with her ovipositor. This would be difficult or impossible for endoparasitic spedes that oviposit their eggs in die body cavity of die host. In contrast, ectoparasitic spedes lay dieir eggs on K- Takaju is now at the Faculty of Agriculture, Kobe University, Kobe 657, Japan. P. J. Ode is now at the Department of Zoology, Ariiona State University, Tempe, AZ USA. Address reprint requests and correspondence to M. R. Strand. Received 10 February 97; accepted 15 April /97/S5.00 C 97 Iniernanonal Society for Beh»rioral Ecology die body of die host, where they are accessible to a second female. While few endoparasitoids practice ovidde, several ectoparasitic wasps exhibit oviddal behavior (Arakawa, 87; Goertzen and Doutt, 75; Hardy and Blackburn, 91; Mayhew, 96; Price, 70). One of diese spedes is Bracon hebetor (Antolin et al., 95; Strand and Godfray, 89), a gregarious, larval ectoparasitoid of stored grain moths such as Plodia interpuncttua. Strand and Godfray (89) developed a game theoretic framework for examining die evolution of oviddal behavior. They found that ovidde will be most frequent under conditions where die competitive advantage of a first dutch is large, die time necessary to kill eggs is short, and encounter rates with parasitized hosts are high. Experimental studies demonstrated that die competitive advantage enjoyed by first dutches of B. hebetor is large and that, on average, females commit ovidde more frequently when host encounter rates decline or when die proportion of parasitized hosts encountered increases (Strand and Godfray, 89). Here, we further examine environmental and genetic determinants of ovicide in B. hebetor by (1) characterizing die temporal costs of ovidde, (2) examining how host quality, foraging experience, and die number of eggs available for oviposition influence oviddal behavior, and (S) determining whether genetic variation in oviddal behavior exists among populations of B. hebetor. GENERAL METHODS Seven different populations of B. hebetor, designated as Wke, PA, Cho, d-, Lum, Bam, and d-bam were used during die study. Wke is a black eyed, wild-type population established from 75 female wasps collected in a corn storage in Waunekee, Wisconsin, USA, in September 90 (Antolin and Strand, 92; Ode et al, 96). Bam originated from wasps collected in Bamberg County, Soudi Carolina, USA, in May 88. O (ivory eyes), Pch (red eyes), and Cho (cantaloupe eye color, honey body) are mutant populations maintained continuously in laboratory culture since die 30s (Whiting, 61). Z-umis a black-eyed, wild-type population maintained continuously in die laboratory since approximately 70 (Petters and Menus, 80). The d-bam culture was established from a cross between die d and Bam cultures as described by Ode et al

2 648 Behavioral Ecology Vol (96). All wasp cultures were maintained using P. mterpunc- Uila as the host at 27 C and a 16:8 h lighcdark cycle (Antolin et al, 95). P. mterptmdtua was reared under identical physical conditions on a diet of cornmeal, poultry feed, and gfycerol (Ode et al, 95). All experiments were conducted at 25 i 1 C in ambient humidity during daylight hours using 6-cm diam plastic petri dishes as assay arenas. In all experiments day 1 indicates the day the adult wasp emerged from its pupal cocoon. All wasps used in the study had head-capsule widths of mm and were mated. Female wasps of this size range have a statistically similar daily fecundity, lifetime fecundity, and longevity when provided continuously with hosts (Ode et al, 96). On each day of an assay, we moved females to a new culture dish that contained either an unparasitized host, a parasitized host, or no host. AD hosts used in experiments were mg, fifth instar P. interpuncteua larvae. Unless stated otherwise, parasitized hosts had first clutches of 8-12 eggs laid h previously by an unrelated, conspedfic female. Females were provided a 50% honey solution as an energy source on days when they were not given a host. We refer to the eggs laid by a female on a previously unparasitized host as the first clutch and the eggs laid by a second, superparasitizing, female as the second dutch. We simulated different rates of host encounter by presenting female wasps on successive days with either: a high-quality, unparasitized host (U); a low-quality, parasitized host (P); or no host (N). As a shorthand method for describing specific experiments, we use P, U and N as abbreviations for what a female encountered on a given day. For example, an experiment in which a newly emerged wasp (day 1) encountered no hosts for S days, a parasitized host on days 4 and 5, and an unparasitized host on day 6 is written as NjP,U. Temporal cost of ovkide Models investigating the dutch size and superparasitism decisions of parasitoids generally assume that reproductive success is limited by either the time available for locating and parasitizing hosts (time-limited spedes) or by the number of eggs available for oviposition (egg-limited spedes) (summarized by Godfray, 94). Parasitoids like B. hebetorcan be considered both time and egg limited. Females require 2-12 h to oviposit depending on host size, encounter rates with hosts, and physiological state (Ode et al., 96; Strand and Godfray, 89; Takasu et al, unpublished data). B. hebttor is also synovigenic, with females producing eggs throughout adulthood (Ulryett, 45; Whiting, 61). To mature its eggs, B. htbetor host feeds on some of die same hosts used for oviposition (Ode et al., 96). Average lifetime fecundity for females provided hosts ad libitum is 614 eggs; ma-rimmri daily fecundity is eggs (Ode et al, 96). However, females begin to reabsorb mature eggs within 3 days of not encountering hosts, even when provided access to honey or sucrose (Benson, 73; Takasu K et al, unpublished data; Ben-Shalom S and Strand MR, unpublished data). Thus, both daily and lifetime egg production are linked to host encounter rate because host feeding is required for egg maturation. The amount of time required to oviposit is also influenced by encounter rate because females with few or no mature eggs will, upon finding a host, remain with it after host feeding until eggs can be matured and laid. In light of these life history characteristics, we assessed the temporal costs of ovidde by characterizing the oviposition behavior of B. hebetot. Methods We randomly selected mated, newly emerged females from the Wke culture and placed them individually into 20-cm plastic culture dishes. Females in one treatment were fed honey for 3 days after emergence and then given a parasitized host on the fourth day (NjP), whereas females in a second treatment were given one unparasitized host per day for 3 days and a panitized host on the fourth day (UjP). A host presentation schedule of N,P simulated a low encounter rate with poor-quality hosts, and U,P simulated a high encounter rate with predominantly high-quality hosts. AD parasitized hosts had first clutches laid by an unrelated Pch female. The position of each egg in the first dutch was noted so that, after oviposition by the second female, all eggs in the first and second clutch could be distinguished from one another. Behavioral observations began when the female encountered die parasitized host and finished when die wasp left the host for more than 20 min or at the end of 8 h. During die intervening period, each female was observed continuously under a stereoscopic microscope and her behavior recorded. Because B. hebetor females lay their eggs underneath the host, we placed an angled mirror beneath the culture dish so that oviddal and egg laying behavior could be observed directly. We recorded behavioral events to the nearest minute. Remits Females in both treatments encountered the parasitized host within 20 min of being placed in die assay arena. Eighty two percent of the females that experienced a low host encounter rate (N,P; n " 17) committed ovidde and laid a second dutch (i.e., superparasitized), whereas only 35% of females (n - 28) that experienced a high encounter rate (U,P) did so (G = 12.1, df = 1, p <.001). All females (n = 3) that did not commit ovidde in the (N,P) treatment laid a second dutch, but only 33% of nonoviddal females laid a second dutch in the (UjP) treatment The remaining nonoviddal females rejected die host after the examination phase of oviposition (see below).. Oviposition was divided into five distinct behavioral phases: stinging, host examination, host feeding, egg laying, and resting (Table 1). Although parasitized hosts were already paralyzed, each female bent her abdomen forward, extended die ovipositor, and briefly inserted die tip of the ovipositor into die host. After several minutes, each female climbed on top of die immobilized host and initiated host examination. Examination was characterized by die female mounting the host and repeatedly walking from one end to die other while antennating and probing the ventral side of die host Wiu the ovipositor. Females almost always contacted one or more eggs in die first dutch with die ovipositor within 5 min of initiating host examination behavior. When this occurred, females diat committed ovidde stopped walking, extended die ovipositor, and punctured all die eggs contacted. Punctured eggs desiccated within 30 min. Females that committed ovidde in both die high and low encounter rate treatment killed more than 80% of die eggs in the first dutch (mean proportion of eggs in die first clutch killed by oviddal females experiencing a low encounter rate = 0.85, n " 14; high encounter rate ** 0.81, n - 8; t = 0.4, df = 20, p>.70, arcsine detransformed data). Indeed, 10 of die 14 oviddal females from die low encounter treatment and 5 of die 8 oviddal females from die high encounter treatment killed all of die eggs in the first clutch. Females that committed ovidde spent, on average, min in die examination phase. We found no evidence that it required less time for females to kill the first egg in a first dutch than to kill other eggs. To die contrary, because eggs are laid in aggregations, once a female located a dutch, she pierced multiple eggs in rapid succession. All females that rejected or superparsitized a host without committing ovidde contacted eggs of die first dutch with their ovipositors during host examination. However, these females moved their ovi-

3 Takasu et al Determinants of ovidde in parasitic wasps 649 Tkbtel Duration of oehsrionu < dated with ovipoaition on pansitlxed hosts Avenge duration (min) ± SD for each behavioral event Treatment and outcome* N Stinging Examination Host-feeding Oviposition Resting* Total Low encounter rate (NJP) Committed ovidde and superparasitized 14 Did not commit ovicide but superparasitized 3 Rejected host 0 High encounter rate (U,P) Committed ovicide and superparasitized 8 Did not commit ovicide but superparasitized 6 Rejected host ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±.2 75 ± ± ± 15 4 ± * ± ^ ± ± ± ± 50.6 * Females encountered i parasitized host after experiencing either a low (N,P) or high (U,P) rate of host encounter. Females in each treatment could either commit ovidde and superparasitize, superparasitize without committing ovidde, or reject the host. b Average resting dme for females reported here includes only the time devoted to this behavior during the 8-h period we observed wasps continuously. Most females who experienced a low rate of host encounter remained in the resting phase through the fouowing day. This yields an estimated total time in the resting phase of > 12 h (720 min). positors away without piercing any eggs and continued examining other parts of the host. Host examination by nonoviddaj females lasted less than 10 min After host examination, all of the females in the N,P treatment fed on the host, whereas only one female from the UjP treatment fed on the host. During host feeding, the female remained on top of the host and drilled through a dorsal region of the cuticle with the ovipositor. After withdrawing the ovipositor, the female placed her mouthparts over the wound and fed on hemoh/mph. Females that superparasitized laid their eggs underneath the host while still standing on its dorsal surface. Second dutches were usually laid in close proximity to thefirst clutch. Females in the N S P treatment laid an average of 2.1 eggs by the end of the 8-h observation period; females in the UsP treatment laid 6.2 eggs. After ovipositing, the female walked off the host and entered what we referred to as the resting phase. During this period, females stood 1-2 mm away from the host and rarely moved. Females in the UsP treatment remained next to the host for approximately 1 h. In contrast, all of the N,P females remained next to the host for the remainder of the 8-h observation period (approximately 6 h). These females were not observed during nighttime h, but most were still next to the host the following morning (day 5). By the end of day 5, females laid, on average, an additional 6 eggs on the host before leaving. Our observations suggest that N,P females matured additional eggs during this resting phase and laid them on the host before departing. Due to the large amount of time spent in the resting phase, females that experienced a low host encounter rate overall spent a significantly longer amount of time with die host than females that experienced a high encounter rate (t * 10.9, df = 29, /xr.0001; Table 1). The large difference in total ovipojitdon rime between the two treatments resulted in differences in the percentage of time spent performing other behaviors associated with oviposition. For example, the average amount of rime spent in the host examination phase by females committing ovicide did not differ between the low and high host encounter rate treatments (/ = 0.7, df «20, /»>.4; Table 1). As a percentage of total oviposition time, however, the examination phase accounted for less than 10% of total oviposition time for ovicidal females in the low host encounter treatment, whereas it accounted for more than 30% of total oviposition time for females in the high encounter rate treatment (i = 7.1, df - 20, /xc.0001). Recognition of self- and cmwpecificady parasitized host* If a parasitoid is likely to rediscover hosts previously parasitized by herself, the risks of destroying her own eggs could counter any advantage of ovicide (Smith and Lessells, 85; Strand and Godfray, 89). Field studies suggest that the parasitized hosts a B. htbetor female encounters while foraging are likely to be a mixture of hosts parasitized by herself (selfparasitized) and hosts parasitized by others (conspedncally parasitized) (Antolin and Strand, 92; Ode et al, 96). The ability to distinguish between self- and conspedfically parasitized hosts has not been investigated for any parasitoid reported to practice ovicide. However, studies with odier parasitoids indicate that some species can recognize self parasitized hosts from hosts parasitized by conspecifks (Hubbard et al, 87; McBrien and Mackauer, 91; van Dijkin et al. 91; Visser, 93). We examined whether B. htbetor females have this ability. Methods Newly emerged and mated Wk* (wild-type, black eyes) strain females were isolated in culture dishes as described previously. Each female was fed honey for 4 days, given a single unparasitized host for 3 consecutive days, and then on day 8 was given a parasitized host (N,U,P). On day 8, half the females encountered the host they had parasitized the previous day. The other females encountered a host parasitized die previous day by an unrelated Pch female. Pairing Wke females with Pch females that carry a recessive red eye color mutation (Whiting, 61) allowed us to unambiguously distinguish progeny from first and second clutches. Wke and Pch wasps readily mate with one another, and previous study revealed no differences in die oviposition responses of Wke females when encountering hosts parasitized by Pch females versus other Wke females (Antolin et al., 95). We considered that B. htbetor could recognize self- and conspedfically parasitized hosts in two ways: females either recognize differences in the

4 650 Behavioral Ecology VoL 8. 6 host or host patch, or they recognize differences in the eggs present on the host To distinguish between these possibilities, we repeated the same experiment described above with the exception that the dutches laid on these hosts were reversed (Le., the eggs laid by the Pch female were transferred to die self-parasitized host and vice versa). Because B. hebttorbys its eggs loosely on the ventral side of paralyzed hosts, eggs can be easily transferred from one host to another using a paintbrush or forceps (Ode et al., 96). Control experiments consisted of picking up the eggs laid by the Pch or experimental female and placing mem back on the same host from whence they came. In these experiments, care was taken to place the eggs in the same distribution as they had been laid originally. Route Only 8% of females who contacted hosts they had parasitized (n " 36) committed ovidde, whereas 27.3% of females who contacted hosts parasitized by a conspedfic (n =» 33) committed ovidde (G = 4.4, df» 1, p<.04). In addition, 69.7% of the females laid a second dutch on hosts parasitized by a conspedfic, while 44.4% of females laid a second dutch on hosts they had parasitized previously (G. «* 4.5, df = 1, p<.04). In the egg-transfer experiments, 11.1% of females committed ovidde when presented a self-parasitized host with a conspedfic (Pch) dutch, whereas 29.2% of females committed ovidde when presented a conspecifically parasitized host with a self-laid dutch (G -4.4, df = 1, fkm). Environmental and physiological determinants of ovidde With the variation in density that occurs in field populations of P. mttrpuncteua (Antolin and Strand, 92; Ode et al., 96), B. htbetor likely encounters varying numbers of unparasitized and parasitized hosts while foraging. Because wasps must feed on hosts to produce eggs, host encounter rate will also influence the number of mature eggs available for oviposition. Thus, egg load, the experience gained by contacting hosts at different rates, or both could influence the decision to commit ovidde and/or superparasitize. We conducted three experiments to assess (1) how host encounter rates influence egg loads of B. htbttor and (2) whether egg load or time between oviposition events is a better predictor of whether females commit ovidde. Methods AH females used in these experiments were fi-om the Y/ke population, and females were observed as described previously. We conducted three experiments that were designed to influence egg load, wasp age, and foraging experience in different ways. In experiment A, females were given a single parasitized host on the day of emergence (P), 4 days after emergence (N^P), or 7 days after emergence (N,P). These females encountered the same number of hosts but did so at different times after emergence. In experiment B, females encountered one host per day beginning at 4 days after emergence. In one treatment, wasps encountered three unparasitized hosts before encountering a parasitized host (NjU,P), while in die other, wasps encountered four parasitized hosts consecutively (NjP,,). These females experienced the same host encounter rate but host quality differed. In experiment C, females encountered a parasitized host 1, 2, or 4 days after experiencing a low or high encounter rate with unparasitized hosts. Three combinations of low and high encounter rate treatments were conducted: low encounter rate (N,UP), high encounter rate (NJUJP); low (N,UNP), high (NJUJNP); and low (N S UN,P), high (NjU,N,P). Here, females experienced different encounter rates. Within each low and high host density combination, females differed in age. However, the time between encountering the last unparasitized host and a parasitized host was the same. Immediately after a female left the last (parasitized) host, we examined whether she committed ovidde, and we counted the number of eggs she laid as a second dutch. A female was scored as oviddal if she killed two or more eggs of the first dutch. As noted previously, however, most females committing ovidde usually killed >80% of the eggs in tile first dutch. The female was then immediately dissected and the mature eggs in her ovaries were counted. Unlaid eggs were scored as mature if they had a chorion and were no longer enveloped by fblh'de cells. The total egg load for each female was the sum of the number of eggs laid on the host phis all mature eggs remaining in the ovaries. Results Ovidde rates, the number of eggs laid by superparasitizing females, and egg loads for experiments A-C are presented in Table 2. In experiment A, most females in each treatment committed ovidde. However, die number of eggs laid and total egg loads both decreased significantly with increasing time between emergence and when die parasitized host was encountered. For example, day 1 females had egg loads of nine, whereas day 7 females had egg loads of less than two. In experiment B, die proportion of females that committed ovidde did not differ significantly between treatments (N,U 5 P and N 3 P 4 ). Females tha.t encountered only parasitized hosts laid larger second dutches and had marginally larger total egg loads than females who had previously encountered unparasitized hosts. In experiment C, the proportion of females committing ovidde increased with die time between when the last unparasitized host was encountered and when the parasitized host was encountered. This increase in rate of ovidde was not influenced by the density of hosts initially experienced by die female. For example, approximately 40% of females committed ovidde the day after experienring a low (NjUP) and high (N,U S P) density of unparasitized hosts. In contrast, more than 70% of females committed ovidde if deprived of hosts for three days after experiencing a low (NJUNJP) and high (N,UjN,P) density of unparasitized hosts. As expected, egg loads were larger for females that experienced high host densities than for females who experienced low host densities. However, females that experienced high host densities laid fewer eggs on the parasitized host than females who experienced low host densities. Egg loads increased in females deprived of a hest for 1 day (N,UP and NJUJNP) but decreased after 4 days (N 3 UNjP and N S U, NjP; Table 2). The results presented in Table 2 were also examined in terms of the specific oviposition decisions made by individual females. In each experiment, wasps could commit ovidde and superparasitire, superparasitize without committing ovidde or reject the host (Table 3). Within each experiment, oviddal females consiuendy laid larger second dutches than nonoviddal females. However, egg loads did not differ between females that committed ovidde and superparasitized versus females that did not commit ovidde but still superparasitized. Regardless of whedier a female committed ovidde, females that superparasitized in experiments B and C had larger egg loads than females who rejected hosts. Genetic variation in ovidde Widi die possible exceptions of sex ratio (Antolin, 92; Orzack and Gladstone, 94), superparaiitism (Wanjberg et al., 89), and resistance to hest defense responses (Henter, 95; Kraaijeveld and van Alphen, 94; Mollema, 91), relatively little is known about the genetics of traits associated with host selection by parasitoids. Previous theoretical and empirical explorations suggested that genetic differences in oviddal be-

5 Takasu et al Determinants of ovicide in parasitic wasps 651 Table Z Rates of ovicide, nmnben of eggs bid in a second #*i«i4i, and total egg load* of female* that (A) < I paiasluied h. 1,4, and 7 days after emergence, (B) encountered boats of different quality bat at the same rate, and (C) encountered hosts of different quality at different rates Treatment N Proportion of ovicidal females Number of eggs laid (± SE) Egg load (SE) Experiment A lday(p) 4day(N,P) 7day(N,P) Test ttariffir Experiment B Low-quality hosts (N,P 4 ) Hlgh^uality hosts (N S U,P) Experiment C 1 day after encounter (N,UP) (NJUJP) 2 days after encounter (N.UNP) (N,U,NP) 4 days after encounter (N S UN,P) (N S U^J,P) G , df - 2, p- A G , df - 1, p-m Host density: G , df - 1, p-.70 Days without hosts: G- 11J, df - 2, p<.005 havior could exist among populations of B. htbttor. First, Strand and Godfray (89) predicted that alternative evolutionarily stable strategies may occur in species that regularly encounter parasitized hosts during their lifetime; resulting in populations composed predominantly of nonoviddal or ovicidal female*. Second, the experiments of Antolin et al (95) suggested that different ovipodtional strategies may exist in A htb tor because some females rarely commited ovicide under conditions in which other females regularly commited ovicide. Given that the previous experiments of this study focused on how environmental factors affect ovicidal behavior of Y/kt females, we sought evidence for whether genetic variation in ovicidal behavior exists by comparing females from each B. htbttor population maintained in our laboratory. Methods We compared ovicidal behavior among populations under low rates of host encounter that favored ovicide in Wkt females and high rates of host encounter that did not In the low encounter rate treatment, females were presented with one parasitized host (P) every third day for 7 days (PNJPNJP). In the high encounter rate treatment, females were presented a parasitized host (P) every third day and an unparasitized host (U) during intervening days (PUjPU,?). The proportion of females committing ovicide was scored as described previous i * 0.4 f t- " 13.0 p< J i <*-3.7 p< ± i * i ± i 0.8 Host density: *i.i» p<.10 Days without hosts: Fun - 90 />< ± ± ± 0.4 F %tt - 43J < ± ± 0.8 ^-1.9 p" ± J ± ± ± ± 0.8 Host density: F UM p<.001 Dayi without hosts: F vn <.001 ResuUs Regardless of population, most B. htbttor females committed ovicide on day 1 (Table 4), and rates of ovicide thereafter tended to decline upon subsequent encounters with parasitized hosts (C. = 174.1, df = 2, p <.0001). Also as observed in previous experiments, the proportion of females committing ovicide was higher when females experienced a low rate of host encounter than when females experienced a high rate of host encounter (G = 4.74, df «1, p <.03). However, we detected significant variation among populations in overall rates of ovicide (G.» 77.8, df 6, p <.001). Among population differences were also detected in how much ovicide decreased after each successive encounter with a parasitized host (strain-by-day interaction: G» 2S.6, df - 12, p <.03). For example, females from the d, Bam, and 6-Bam populations usually committed ovicide with each encounter of a parasitized host in both the low and high encounter rate treatment (Table 4). In contrast, females from the Cho and Pch populations rarely committed ovicide after the first encounter with a parasitized host. The proportion of females that committed ovicide decreased with encounter rate in each population (population by encounter rate interaction: G = 6.86, df m 6, p >.33). DISCUSSION The benefit of ovicide is that it allows a parasitoid to increase the value of a parasitized host by eliminating some or all of the progeny in a first clutch. The potential costs of ovicide include (1) the time required to kill eggs could reduce the time available for other activities such as foraging for new hosts, and (2) the risks of killing one's own eggs could eliminate any advantage of ovicide if females cannot Hinringirish between self- and conspeafically parasitized hosts. Previously, we established that first dutches of B. htbttor have a substan-

6 652 Behavioral Ecology Vol T«bte3 Ovliwfaioa dedsfons, sin of FTT^IM? «IHI ii»f laid, T«H for females that (A) encountered parasitised boats 1, 4, and 7 days after emergence, (B) encountered hosts of different quality at the same rate, and (C)encotuMened hosts at different rates* Committed ovicide Experiment A Experiment B Experiment C Superparasitized N Number of egg* laid(±se) _ _ ^IJ. " 11 p> JO 3.7 i i 0.7 F lm " 9.1 p<.006 4J J F lm " 25.2 p<.0001 Egg load (±SE) _ 1.6 i ± ± 0.6 Fi* a 1-9 p >.17 Superparasitized: p'< J i J ± ± 0.9 F, M p> M Superparasitism: F tm = 5.1 p< ± ± ± /,..» " 0.39 p>.90 Superparasiusm: 'us These data are from the same females as presented in Table 2. Females are categorized by whether they committed ovicide, superparasitized, or rejected the host. rial competitive advantage over any eggs laid subsequently on a host and that ovicide enhances the reproductive success of females under conditions where host encounter rates are low and the proportion of parasitized hosts are high (Strand and Godfray, 89). Consistent with the prediction that ovicide is most likely to occur when kill times are short (Strand and Godfray, 89), we found that the temporal cost of ovicide to B. hebetor is relatively small for females that experience low levels of host encounter but increases for females that experience a high rate of encounter. B. hebetor females that commit ovicide also tend to kill most or all of the eggs in the first clutch. This too is qualitatively consistent with the prediction that ovicide will be "all or none" if the same amount of time is required to kill an egg regardless of the number of eggs destroyed previously (Strand and Godfray, 89). Interestingly, females that do not commit ovicide almost always contact eggs of the first dutch during host examination. This indicates that the failure to kill eggs is not due to a female's inability to locate them. Our results also indicate that & hebttor females commit ovicide and superparasitize hosts attacked by a conspecific more often than self-parasitized hosts. At this time we do not know how these wasps discriminate between self- and conspedfically parasitized ho*ts,but the cue(s) appears to be associated with the host rather than the eggs laid on the host Many ichneumonoids recognize parasitized hosts via marking pheromones produced by the Dufour's gland (summarized by Vinson, 85), suggesting this as a possible source for a marking pheromone in B. hebttor We also note that avoidance of ovicide on self-parasitized hosts is not perfect; indicating that females face somerisk of killing their own eggs while foraging. Rosenheim and Mangel (94) found in their theoretical explorations that even small error rates in discrimination between self- and conspedfically parasitized hosts will favor early departure from partially exploited host patches. One adaptation by & hebetorto minimize the risks of self-ovidde would appear to be the ability to recognize the time between successive ovipositions (see below). As shown in Table 2, most females commit ovicide if they have never encountered a host or if more than S days have elapsed since last encountering a host. Since B. hebetor eggs batch under standard conditions in 60 h, high rates of ovidde in association with encounter rates of less than one host per 3 days may reflect an evolutionary outcome to minimizing the risks of committing self-ovidde. Given the interrelationship between oviposition history and egg load in B. Ktbetor, the second part of our study examined what factor(s) most % influences the decision to commit ovidde. As expected, rates of superparasitism and ovidde increase when unparatitized hosts are rare and the absolute density of hosts is low. However, our experiments suggest that B. hebetor learns about its environment while foraging and adjusts its decision to commit ovidde on the basis of the elapsed time between successive hosts. Consider the results presented in Tables 2 and 3. At emergence, females have eight to nine mature eggs available for oviposition but egg loads decline rapidly, via reabsorption, if no hosts are encountered. Females almost always commit ovidde and superparasitize if the first host they encounter is parasitized. The frequency of ovidde thereafter declines regardless of host quality so long as hosts are encountered on a daily basis. In contrast, the proportion of females committing ovidde increases if wasps experience successive days without hosts. Changes in host acceptance and dutch size responses have been reported in some parasitoids to be correlated with differences in egg load (Minkenberg et al, 92; Rosenheim and Rosen, 91; Volkl and Mackauer, 90). However, the decision by B. hebetor to commit ovidde and superparasitize is dearly not correlated with this measure of physiological state. Compare, for example, females in experiment A that did not encounter a host until day 7 (NgP) with females from experiment C that encountered one unparasitized host before encountering a parasitized host on day 8 (N 3 UN,P). Females in both treatments were of similar age and usually committed ovidde, yet in experiment A females had egg loads <2, while in experiment C females had egg loads >9. Although wasp age was not held constant in these experiments, the overall trends strongly suggest that the interval between host encounters is the most important determinant of oviddal behavior in this species. Since B. hebetor usually commits ovidde when the first host encountered is parasitized, we hypothesize that females begin life estimating there is no risk of committing self-ovidde and that host densities are low. Females thereafter adjust their estimate for future reproductive success and avoidance of selfovidde as they forage in the environment This pattern parallels most strongly dynamic state variable models whereby organisms are predicted to make reproductive decisions on the basis of their "informational sate" (e.g., Mangel and Clark, 88). A widespread observation among gregarious parasitoids is that larger dutches are laid on higher quality, unpar-

7 Takasu et al Determinant! of ovicide in parasitic wasps 653 Tkble4 Proportion of females who roninilltnl ovidde from seven popul i of B. htbttor tested m a low boat density, low host quality environment (PNJPNJP) and a Ugh host density, high host quality emuumumt (PUJPUJP) Treatment Day parasitized host encountered Proportion of females in each population committing ovidde Bam &-Bam V/ki Lum Oe Pch Total Low host density (PN.PN.P) N High host density (PU.PU.P) N J asitized hosts than on lower quality, parasitized hosts (van Dijken and Waage, 87; Wyiie, 65). We too find that B. hebetor females who commit ovidde consistently lay larger second clutches than females who superparuitize without committing ovicide. This suggests that females rank parasitized hosts following ovicide as being of higher quality than parasitized hosts with an intact first clutch. However, we find it surprising that females superparasitize some hosts without committing ovicide considering the large competitive advantage of a first dutch and the relatively small temporal costs of this behavior. Dynamic state variable approaches based on the parameters developed here could clarify how superparasitism without ovidde might be favored. Strand and Godfray (89) predicted that oviridal and nonoviddal strategies could evolve under conditions where superparasitism is common. Our own field and laboratory studies suggest that B. hebetor experiences foraging conditions that meet this criterion. In the field, wasp populations can build to high levels where competition among females for hosts is intense (Antolin and Strand, 92; Ode et al, 96). Our results indicate that rates of ovidde vary significantly among populations, although at this time we can only speculate as to whether drift, inadvertant selection, or some combination thereof is responsible for these differences. Regardless, these results indicate that oviddal behavior has a genetic component, that environmental factors may influence oviddal behavior differently across populations of B. hebetor, and that this behavioral trait may be under selection in the field. At this time we do not know what form of selection might maintain oviddal and nonoviddal individuals within a population. One possibility is that variation is maintained via density-dependent selection whereby ovidde becomes favored as parasitoid densities and competition for host resources increase during the late summer and autumn. Alternatively, variation could be maintained at intermediate levels by frequency-dependent selection if ovidde is costly in terms of future reproductive success as might occur when host densities are high. Finally, variation in oviddal behavior among populations may reflect selection on genetically correlated traits assodated with oviposition. How environmental and genetic components interact to affect oviddal behavior in field populations of B. hebetor is currently under investigation. We thank S. Zimmers and J. A. Johnson for assistance in rearing of wasps and hosts, and G. Heimpel, M. Mangel, and two anonymous reviewers for their critical reading of the manuscript. This work was supported by U.S. Department of Agriculture grant and University of Wisconsin Hatch Project REFERENCES Antolin MF, 92. Sex ratio variation in a parasitic wasp. II. Diallel cross. Evolution 46: Antolin MF. Strand MR, 92. Mating system of Braam htbttor (Hymenoptera: Braconidae). Ecol Entomol 17:1-7. Antolin MF, Ode PJ, Strand MR, 95. Variable sex ratios and ovicide in an outbreeding' parasitic wasp. Anira Behav Arakawa R, 87. Attack on the parasitized host by a primary solitary parasitoid Encania formosa (Hymenoptera: Aphelinidae): the second female pierces with her ovipositor, the egg laid by the first one. Appl Entomol Zool 22: Benson JF, 73. Intraspecific competition in the population dynamics of Braam hebttor Say (Hymenoptera: Braconidae). J Anim Ecol 42: Chamov EL, Skinner SW, 88. Clutch size in parasitoids: the egg production rate as a constraint. Evol Ecol 2: Godfray HCJ, 94. Parasitoids: behavioral and evolutionary ecology. Princeton, New Jersey: Princeton University Press. Goertzen R, Doutt RL, 75. The ovicidal propensity of Gomozus. Ann Entomol Soc Am 68: Hardy ICW, Blackburn TM. 91. Brood guarding in a bethylid wasp. Ecol Entoraol 16: Henter HJ, 95. The potential for cocvolution in a hosi-paraatoid system. II. Genetic variation within a population of wasps in the ability to parasitize an aphid host. Evolution 49: Hubbard SF, Marris GC, Reynolds A, Rowe GW, 87. Adaptive pat terns in the avoidance of superparasitism by solitary parasitic wasps. J Anim Ecol 56: Iwasa Y, Suzuki Y, Matsuda H, 84. Theory of oviporition strategy of parasitoids. I. Effect of mortality and limited egg number. Theor Popul Biol 26: Kraaijeveld AR, van Aiphen JM, 94. Foraging behavior and encapsulation ability of Drxaophiia mdanogasur larvae: correlated polymorphisms (Diptera: Drosophilidae)?J. Insect Behav 8: Mangel M, 92. Descriptions of superparasitism by optimal foraging theory, evolutionarily stable strategies and quantitative genetics. Evol Ecol 6: Mangel M, Clark CW, 88. Dynamic modeling in behavioral ecology. Princeton, New Jersey: Princeton University Press. Mangel M, Rosenheim JA, Adler FR, 94. Clutch size, offspring performance, and intergenerabonal fitness. Behav Ecol 5: Mayhew P, 96. Ecological studies of insect reproductive behaviour (PhD thesis). London: University of London. McBrien H, Mackauer M, 91. Decision to superparasitise based on larval survival: competition between aphid parasitoids: Aphidha trvi and ApJndius smithi. Entoraol Exp Appl 59: Minkenberg OPJM, Tatar M, Rosenheira JM, 92. Egg load as a major source of variability in insect ovipoiition behavior. Oikos 65: MollemaC, 91. Heritability estimates of host selection beahviour by the Drotophtia parasitoid Atobara tabida. Neth J Zool 41: Ode PJ, Antolin MF, Strand MR, 95. Brood-mate avoidance in the parasitic wasp, Braam htbttorsxy. Anim Behav 49:

8 654 Behavioral Ecology Vol Ode PJ, Antolin MF, Strand MR, 96. Sex allocation and sexual asymmetries in intra-brood competition in the parasitic wasp Bracon htbttmj Anlm Ecol 65: Orzack SH, Gladstone J, 94. Quantitative genetics of sex ratio traits in the parasitic wasp, Naionia vitripmnis. Genetics 157: Petters RM, Menus RV, 80. Decreased diploid male viability in the parasitic wasp, Bracon ktbttotj Hered 7135S-356. Price PW, 70. Biology and host exploitation by PUolofJms mddstmcfun (Hymenoptera: Ichneumonidae). Ann Entomol Soc Am 63: Roitberg BD, Mangel M, Lalonde RG, Roitberg CA, van Alphen JJM, Vet L, 92. Seasonal dynamic shifts in patch exploitation by parasitic wasps. Behav Ecol 3: Rosenheim JA, Rosen D, 91. Foraging and oviposition decisions in the paradtoidaphytis Hngnantnsic distinguishing the influences of egg load and experience. J Anlm Ecol 60: Rosenheim JA, Mangel M, 94. Patch-leaving rules for parasitoids with imperfect host discrimination. Ecol Entomol Skinner SW, 85. Clutch size as an optimal foraging problem for insects. Behav Ecol Sociobiol 17: Smith RH, Lessells CM, 85. Oviposition. ovicide and lanral competition in granivorous insects. In: Behavioural ecology (SiWey RM, Smith R, eds). BlackweU, Oxford; Strand MR, Godfray HCJ, 89. Superparasitism and ovicide in parasitic Hymenoptera: theory and a case study of the ectoparasitoid Bracon htbttm Behav Ecol Sociobiol 24: Ullyett GC, 45. Distribution of progeny by Micnbnuo* httttorsxy. J Entomol S Africa 8: van Alphen JJM, Viiaer ME, 90. Superparasitism as an adaptive strategy for insect parasitoids. Annu Rev Entomol 35: van Dijken MJ, Waage JK, 87. Self and conspecific superparasitisni by the egg parasitoid Tridiograwano tvanttans. Entomol Exp Appl 43: van Dijken MJ, Neuenschwander P, van Alphen JJM.Hammond WNO, 91. Sex ratios in field populations of Epidinocamt lopax, an exotic parasitoid of the cassava mealybug in Africa. Ecol Entomol 16: Vinion SB, 85. The behavior of parasitoids. In Comprehensive insect physiology, biochemistry and pharmacology (Kerkut GA, Gilbert LI, eds). Oxford: Pergamon Press; Visser ME, 93. Adaptive self- and conspedfic superparasitism in the solitary parasitoid Ltptopi&na hturotoma (Hymenoptera: Eucoilidae). Behav Ecol 4: Vuaer ME, van Alphen JJM, Hemerik L, 92. Adaptive superparasitism and patch time allocation in solitary parasitoids: an ESS model. JAnim Ecol \folkl W, Mackauer M, 90. Age-tpecinc pattern of host discrimination by the aphid parasitoid Ephtttms catifomicus (Hymenoptera: Aphidlidae). Can Entomol Wanjberg E, Pizzol J, Fouillet P, 89. Genetic variation in progeny allocation in Trichograwtwia maidis. Entomol Exp Appl 53: Weisser WW, Houston AI, 93. Host discrimination in parasitic wasps: when b it advantageous? Funct Ecol 7: Whiting AR, 61. Genetics of Habrobracon. Adv Genet 10: Wylie HC, 65. Discrimination between parasitized and unpantiuzed housefly pupae by females of Naionia vitrtpmnii (Walker) (Hymenoptera: Pteromalidae). Can Entomol 97:2/9-286.