Behavioral Ecology Vol. 7 No. 3: 235-242 Functional shifts in the use of parasitized hosts by a tephritid fly: the role of host quality Daniel R. Papaj" and Russell H. Messing15 a Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ 85721, USA, and Department of Entomology, University of Hawaii, Honolulu, HI 96822, USA b Superparasitism, a phenomenon in which parasitic insects lay eggs in already-exploited hosts, provides a useful context in which to examine the dynamics of parental investment. This study explored conditions under which female Mediterranean fruit flies (Ceratitis capitata) shift from avoiding superparasitism of host fruit to preferring it, even placing eggs direcdy into existing egglaying cavities. An a priori hypothesis of costs and benefits was used to predict how use and avoidance of parasitized fruit would change in -response to changes in fruit size and ripeness. We predicted that avoidance would decrease with increasing fruit size, while use would increase with decreasing ripeness. Using afield-cageassay, ripeness was held constant and the size of host coffee berries manipulated. Avoidance of parasitized berries was significandy less pronounced on large berries than on small ones. In a second experiment, size was held constant and ripeness manipulated. On unripe berries, females deposited the majority of clutches direcdy into existing egg-laying cavities. On ripe berries, by contrast, die same females deposited most clutches in previously unparasitized fruit. Parallel patterns in the frequency of female-female contests were observed, supporting die nodon diat a fruit's value is determined by an interaction between fruit size or ripeness, on one hand, and die prior occurrence of eggs, on die odier. Laboratory assays suggested that use of existing sites had advantages in terms of time savings; female behavior thus constitutes a relatively uncommon example of adaptive superparasitism in which parasitized hosts are actually preferred over unparasitized ones. Key words: Ceratitis capitata, competition, host-marking pheromone, parent-offspring conflict, parental investment, superparasitism, Tephritidae. [Behav Ecol 7:235—242 (1996)] S hifts in patterns of parental investment in animal populations, particularly under real-world conditions, have been poorly documented (Clutton-Brock, 1991). Recendy, such shifts have been examined intensively widi respect to die occurrence of superparasitism, a phenomenon in which parasitic insects lay eggs in hosts already parasitized. For example, females in a number of phytophagous and entomophagous species avoid laying eggs at sites diat have been exploited by conspecifics (reviews by Godfray, 1994; Papaj, 1993; Roitberg and Prokopy, 1987), a behavior diat minimizes die level of intraspecific competition incurred by a female's progeny. Avoidance of previously exploited hosts is usually mediated by a chemical mark, termed a host-marking pheromone, put down by die ovipositing female on or in the host (reviews by Roitberg and Prokopy, 1987). Given costs to die female in terms of pheromone production and additional search effort, avoidance of parasitized hosts appears to represent a legitimate (albeit simple) form of maternal investment in die interests of her offspring. Avoidance of superparasitism, as die deposition of eggs in parasitized hosts is often termed, is ideal for die analysis of maternal investment because it is easily quantified and has clear fitness consequences. Superparasitism is not avoided absolutely by many parasitic insects. A number of factors including egg load, nutritional status, presence of conspecifics, and even weather conditions influence the extent to which females of a given parasite species avoid use of parasitized hosts (Godfray, 1994; Papaj, 1993; Roitberg et al., 1992, 1993; Visser et al., 1990). Recent dieoretical effort indicates that superparasitism can at least sometimes be adaptive, a perspective diat has received significant empirical support (van Alphen and Visser, 1990; Visser et al., 1990). For example, if unparasitized hosts are rare and dierefore hard to find, it can benefit a female to use parasitized Received 27 December 1994; revised 17 July 1995; accepted 13 August 1995. 1045-2249/96/S5.00 © 1996 International Society for Behavioral Ecology hosts. Consistent widi diis prediction, female tephritid fruit flies progressively deprived of die opportunity to lay eggs avoid parasitized host fruit to progressively lesser degrees (Roitberg and Prokopy, 1983). Notwidistanding die adaptiveness of superparasitism under certain conditions, a parasitized host has almost always been presumed to represent a relatively low-quality resource to a female. Given a simultaneous choice between parasitized and unparasitized hosts similar in all odier respects, a female is presumed to prefer to lay eggs in die unparasitized host (van Alphen and Visser, 1990). This presumption is implicit in models of die evolution of host-marking pheromone communication systems (review by Godfray 1994; Roitberg and Mangel, 1987) and, to date, has been supported by behavioral data for a variety of insect species (review by Papaj, 1993). In this context, the behavior of die Mediterranean fruit fly, Ceralilis capitata (Diptera: Tephritidae), is particularly intriguing. As in taxonomically related tephritid species, the Mediterranean fruit fly (hereafter referred to as medfly) marks die fruit after egg laying. As in those species, the host-marking pheromone (when applied experimentally to intact fruit and tested against an unmarked control) has a deterrent effect on egg laying (Papaj et al., 1989a, 1992; Prokopy et al., 1978). Since larvae are constrained to complete development widiin the fruit in which they are deposited, the deterrency of hostmarking pheromone presumably functions to reduce die level of competition incurred by a female's young. This inference is supported by abundant evidence that medfly larvae compete within fruit (Debouzie, 1977a,b, 1978, 1981; Papaj etal., 1989b). Yet despite die deterrency of host-marking pheromone and evidence of its functional significance, medfly females sometimes prefer to lay eggs in egg-infested hosts, often depositing clutches direcdy into egg-laying punctures established by anodier female (Back and Pemberton, 1915; Papaj etal., 1989a, 1992; Papaj, 1993 and references within; Silvestri, 1914). Use of existing egg-laying sites has also been documented in walnut-infesting Rhagolelis flies (Lalonde and Man- Behavioral Ecology Vol. 7 No. 3 236 gel, 1994; Papaj, 1993, 1994b) and several entomophagous parasitoids (review by Godfray, 1994; Price, 1970; Takasu and Hirose, 1988). Deposition of clutches into existing egg-laying cavities, as observed in medfly and walnut flies, holds several possible advantages for females (Lalonde and Mangel, 1994; Papaj, 1993, 1994). In medfly, use of existing sites is associated with increases in the proportion of egg-laying attempts that result in successful deposition of a clutch (Papaj etal., 1992) as well as a reduction in the time required to deposit a clutch (Papaj, 1993). If females are time-stressed (cf. Papaj, 1993), time savings might translate to increases in the rate at which eggs are laid. Although firm evidence on this point is lacking, use of existing sites may also reduce ovipositor wear and improve the success with which a female is able to establish her own sites (Lalonde and Mangel, 1994; Papaj, 1993, unpublished data). In short, we expect females to use existing sites when the benefits of doing so in terms of time savings or reduced ovipositor wear exceed costs in terms of added competition incurred by their young. In the following study, we were particularly interested to know if a female medfly's use of existing sites might change from a preference for such sites to avoidance of them in accordance with changes in perceived costs and benefits. Such a finding would amount to a demonstration that, under one set of environmental conditions, females act overtly in their offspring's interests and avoid superparasitism, while under other conditions, females act overtly against those interests and superparasitize. In our study, we took advantage of the fact that it is relatively easy to manipulate the fly's host environment and to observe host foraging behavior under conditions closely approximating natural ones. We employed an experimental approach to the dynamics of host use in which the presumed costs and benefits of using existing egg-laying sites were manipulated in realistic ways and a priori predictions about female behavior evaluated directly. METHODS Experimental rationale and predicted effects We first set out to show that females behave on test fruit as though larvae in successive clutches compete with one another. We were not interested in quantifying larval competition; rather, we were interested in manipulating factors known to affect levels of larval competition in tephritid flies and assaying female response to these manipulations. The presumed cost of larval competition associated with use of parasitized fruit was manipulated experimentally by varying fruit size. In another RhagoUlis species, R. pomonella, small fruit provide less substrate for developing tephritid fly larvae than large fruit and are associated with correspondingly higher levels of competition at a given larval density (Averill and Prokopy, 1987a). We predicted that females foraging on large fruit would avoid laying eggs in parasitized fruit to a lesser degree than females on small fruit. We next set out to show that, despite a capacity to act to reduce effects of larval competition, females sometimes act to increase levels of competition by using existing sites. The presumed benefits of use of existing sites associated with time savings and/or ovipositor wear were manipulated by varying fruit penetrability or, equivalendy, fruit ripeness. On a ripe fruit whose soft surface is penetrated quickly by a female's ovipositor, the success in depositing clutches should be relatively high, the time saved by exploiting existing punctures relatively low, and ovipositor wear suffered by establishing new sites relatively slight (Papaj, 1993; Papaj et al., 1992). For all of these reasons, we predicted that females foraging on ripe fruit would use existing sites to a lesser degree than females on unripe fruit. General protocol Experiments were conducted on the grounds of the Kauai Research Station in Kapa'a, Hawaii. All flies used in tests were of wild origin, collected as larvae in loquats from the island of Maui, and held in 25 cm X 25 cm X 25 cm cages at ambient temperatures and photoperiod. All assays were conducted using previously unparasitized coffee berries (Coffea arabica, Rubiaceae) collected from an unmanaged grove near Kapa'a. Coffee was used because, of the over 250 species of fruits, nuts, and vegetables attacked by medfly over its currendy cosmopolitan distribution, coffee is one of the hosts used by medfly widiin its original range in eastern and southern Africa (Christenson and Foote, 1960; Hagen et al., 1981; Liquido et al., 1991). Coffee is also used in the Hawaiian islands from which our flies were collected, and fruit was readily available for our experiments. A covered, cylindrical, nylon-screen field cage measuring 3 m high and 3 m in diameter was placed over a small tree located in the station arboretum. At the beginning of each experiment, pairs of parasitized and unparasitized coffee berries were distributed uniformly throughout the foliage. Berries were hung from branches with the aid of small metal clips and wire threaded through the stem. In this way, we avoided damaging the fruit surface in a way that might influence fly behavior. We then introduced approximately 50 previously mated medfly females of wild origin (age approximately 14 to 21 days) into the field cage and permitted them to forage in the trees for host fruit Observations of fly activity on a particular fruit were made as follows. The foliage was surveyed until a female was observed on a berry. The pair of berries to which this berry belonged became a focal berry pair on which fly activity was monitored until such time as no flies remained on the fruit or 10 min had elapsed. In this way, a record was obtained not only of the first fly noted on a member of the pair, but also of any flies that landed subsequendy on either berry until the end of an observation. For each fly, all oviposition attempts, host-marking bouts, encounters, and contests with other females were recorded with the aid of an audio cassette recorder. Oviposition attempts are characterized by the insertion of the ovipositor tip into the fruit followed by a stereotyped boring behavior. Host-marking bouts (which are unambiguous indications of clutch deposition; Prokopy et al., 1978) are characterized by dragging the ovipositor over the surface of the fruit, usually for 1 to 3 min. Contests were characterized by head-to-head confrontations in which flies approach one another, often with wings folded back. Typically, one fly lunges at the other, often butting the other fly with its head. Most contests end with one of the flies leaving the fruit. Encounters were defined arbitrarily as occurring when twoflieswere within about 0.5 cm of one another and when at least one member of the pair was facing die other. Any berries in which clutches were deposited by test flies were removed at the end of the observation and replaced with fresh ones. Between the two test days, flies remained in the field cage, but berries were removed. Flies were supplied at all time with food in the form of sucrose and protein hydrolysate placed on a small petri dish positioned within the foliage. Water was supplied in \ials with cotton wicks distributed throughout the foliage. Effect of fruit size In this field-cage experiment, females were permitted to forage freely for either large (mean fresh weight = 1.91 g, SE = 0.033, N = 62) or small (mean fresh weight = 0.90 g, SE = 0.023, N = 69) ripe coffee berries. The mean fresh weight for Papaj and Messing • Dynamics of superparasitism in fruit flies a haphazardly collected sample of ripe coffee berries from the same grove over the same time period, by comparison, was 1.34 g (SE = 0.032, N = 240). Within each size class, fruit were of two types: half of the berries were unmanipulated and unparasitized (i.e., control fruit) and half were egg parasitized and marked. Test berries were infested with eggs and marked by permitting females in a laboratory cage access to previously unparasitized berries. After a female bored into the fruit, deposited a clutch of eggs, and marked the fruit by dragging her ovipositor over the fruit surface, a small adhesive tag was placed over the site. Another female was then permitted to lay eggs at a new site and to mark the fruit The berry was then removed, the tag removed, and the twice-infested, twicemarked fruit used in assays that same day. In each test, 15 pairs of unparasitized and parasitized berries of a particular size class were used at a time. Females were presented first with large berries and then with small berries in successive trials on a given day. The order in which berries of each size were presented to females was alternated on successive days. Focal fruit observations were made as described above. Effect of fruit ripeness The second field-cage experiment was similar to the first except that we manipulated berry ripeness rather than size. Ripeness was judged by color: ripe berries are uniformly dark red in coloration, whereas unripe berries are mainly green with occasional streaks of red. All berries were intermediate in size to the large and small berries used in the previous experiment Twice-parasitized berries were prepared as described above, except that egg-laying punctures were circled with ballpoint pen so that reuse of punctures could be more easily confirmed. As a control, circles of similar size were drawn on unparasitized berries. After sufficient test fruit were generated, 30 pairs of berries of a particular ripeness class were distributed throughout the foliage. On day 1, females were presented in the morning with unripe berries and in the afternoon with ripe berries in successive trials. On day 2, females were presented in the morning with ripe berries and in the afternoon with unripe berries in successive trials. Focal fruit observations were made as described above. Advantages of using existing oviposition sites In laboratory assays, we investigated possible advantages of using existing oviposition sites for fruit of differing size or ripeness. In a small screen cage (25 cm on a side), individual females were permitted to deposit eggs in berries into which two clutches had been deposited previously at separate sites. We counted the number of egg-laying attempts and clutches laid at new sites versus existing sites on the fruit. Deposition of a clutch was inferred from the occurrence of host-marking (Prokopy et al., 1978). We also recorded the time required to deposit a clutch at new versus existing sites. Because, in an earlier study (Papaj, 1993), clutch size did not vary between new and existing sites on the same fruit, a difference in the time of clutch deposition was taken here to reflect a difference in the time required to penetrate the fruit surface and begin deposition of eggs (i.e., setup time). Statistical analysis In field cage experiments, our focal berry pair procedure guaranteed that fruit in both treatments in a given experiment were observed for the same overall amount oftime.The relatively short tenure on fruit combined with the relatively long time between observations of the same fruit pair made it unlikely that repeated observations on a focal pair involved 237 (a) 50 40 0) 30 | 20 O x 10 with eggs without eggs Small Fruit with eggs without eggs Large Fruit (b) E 120 I 5 s 90 60 30 HI 6 0 with eggs without eggs Small Fruit with eggs without eggs Large Fruit (c) with eggs without eggs Small Fruit with eggs without eggs Large Fruit Figure 1 Effects of fruit size on use of egg-occupied versus unoccupied fruit (a) number of clutches laid, (b) number of egg-laying attempts, and (c) percent egg-laying attempts that result in egg deposition. the same individual flies. Pseudoreplication in this respect was thus not a major factor in our experiments. For this reason and owing to the frequent occurrence of zero data for one member of a focal pair, we felt justified in grouping data over focal fruit within a treatment Because data for some variables were relatively scarce and because no obvious effects of test day were detected, we further grouped data within a treatment over all test days. Resulting frequencies were analyzed with log-linear models (Sokal and Rohlf, 1981; SAS, 1985). RESULTS Effect of fruit size Regardless of fruit size, females preferred to deposit clutches in unparasitized fruit (Figure la). Nevertheless, the degree to 238 Behavioral Ecology Vol. 7 No. 3 Table 1 Results of log-linear analysis of fruit size and level of infestation as sources of variation in female behavior on fruit Source df Table 2 Frequency of female-female encounters and contests on coffee berries of different sizes or ripeness levels and different levels of infestation p value Propensity to lay eggs Clutch Size Clutch X size Infest Clutch X infest Size X infest Clutch X size X infest 18.12 0.11 0.44 27.65 7.26 32.92 0.80 <.0001 .74 .50 <.0001 .007 <.0001 .3700 Propensity to engage in contests Contests Size Contests X size Infest Contests X infest Size X infest Contests X size X infest 24.48 0.66 0.37 11.30 1.49 11.30 1.49 <.0O01 .42 .54 .0008 .22 .0008 .22 Type of berry Level of infestation Number of encounters Number of contests Encounters with contests (%) Large Parasitized Unparasitized Parasitized Unparasitized Parasitized Unparasitized Parasitized Unparasitized 30 30 15 79 15 37 57 36 23 23 9 64 5 30 47 15 77 77 60 81 33 83 82 42 Small Ripe Unripe which parasitized fruit were avoided depended on berry size; females deposited relatively fewer clutches in parasitized berries when berries were small (Figure la; G test, G = 9.38, df = 1, p < .005). Fruit size influenced clutch deposition in two ways. First, size influenced the number of attempts initiated into parasitized versus unparasitized fruit (Figure lb). When fruit were small, many more attempts were initiated into unparasitized berries than punctured, marked ones; when fruit were large, by contrast, slightly more attempts were intitiated into punctured, marked berries than unparasitized ones. Second, size influenced the frequency of egg-laying attempts that resulted in egg deposition (Figure lc). When fruit were large, proportionally more attempts resulted in egg deposition on unparasitized berries than on punctured, marked ones; when fruit were small, the pattern was in the same direction but less pronounced. These patterns were evaluated with a log-linear analysis of the following factors: clutch deposition given attempt (yes versus no), infestation level (parasitized versus unparasitized), and size (large versus small). Trie interaction among all three factors was not significant (Table 1), indicating that the manner in which level of infestation affected egg-laying success did not depend significantly on berry size. Nevertheless, several of the two-way terms were significant. There was a highly significant size X infestation term (Table 1) indicating that, as noted above, the overall number of oviposition attempts on parasitized versus unparasitized fruit depended significantly on berry size. In addition, there was a significant clutch deposition X infestation term (Table 1) indicating that, as noted above, the propensity of females to deposit eggs given initiation of oviposition depended on level of infestation. Patterns in female-female interactions On large fruit, the frequency of female-female encounters as well as the proportion of encounters that involved a contest was identical on parasitized and unparasitized berries (Table 2). On small fruit, by contrast, encounters among females were far more frequent on unparasitized fruit. Moreover, on small fruit, the proportion of encounters on unparasitized fruit that involved a contest exceeded that on parasitized fruit (Table 2). These patterns were evaluated with a log-linear analysis of three factors: occurrence of contest given encounter (yes ver- • Also shown is the percentage of encounters that involve a contest. sus no), infestation level (parasitized versus unparasitized), and size (large versus small). The interaction among all three factors was not significant (Table 1), indicating that the manner in which previous infestation influenced the occurrence of a contest did not depend significantly on berry size. However, there was a highly significant size X infestation term (Table 1) indicating that, as noted above, the relative frequency of female-female encounters on parasitized versus unparasitized fruit depended on fruit size. Effect of fruit ripeness Use of parasitized fruit and, particularly, existing sites on those fruit depended critically on berry ripeness. When fruit were ripe, clutches were deposited mainly in unparasitized berries (Figure 2a). When fruit were unripe, by contrast, all but one clutch was deposited in punctured, marked berries (Figure 2a; G test, G = 67.02, p < .0001). Ripeness influenced clutch deposition in two ways. First, ripeness influenced the number of attempts initiated into parasitized versus unparasitized fruit (Figure 2b). When fruit were ripe, more attempts were initiated into unparasitized berries than parasitized ones; when fruit were unripe, by contrast, more attempts were intitiated into parasitized berries than unparasitized ones. Second, ripeness influenced the tendency for an oviposition attempt to result in egg deposition (Figure 2c). When fruit were ripe, proportionally more attempts on unparasitized berries resulted in egg deposition than on parasitized berries. When fruit were unripe, by contrast, just the opposite was true. Proportionally more attempts on parasitized berries resulted in egg deposition than on unparasitized berries. These inferences were supported by log-linear analysis of three factors: clutch deposition given attempt (yes versus no), infestation (parasitized versus unparasitized), and ripeness (unripe versus ripe). There was a significant interaction among the three factors (Table 3), indicating that the manner in which level of infestation affected success of egg deposition depended significantly on the ripeness of the berries. There was also a significant ripeness X infestation term (Table 3) indicating that, as noted above, the relative frequency of attempts on parasitized and unparasitized fruit depended on ripeness. Patterns in puncture use All clutches deposited in unripe parasitized berries were deposited in punctures (Figure 3a). Moreover, the preponderance of egg-laying attempts on such berries were initiated in 239 Papaj and Messing • Dynamics of superparasitism in fruit flies (a) 50 40 TableS Results of log-linear analysis of fruit ripeness and level of infestation as sources of variation in female behavioral responses on fruit Source df Propensity to lay eggs Clutch Ripeness Clutch X ripeness Infest Clutch X infest Ripeness X infest Clutch X ripeness X infest 1 1 1 1 1 1 1 42.18 0.02 12.29 0.53 1.43 35.55 13.19 <.0001 .89 .0005 .47 .23 <.0001 .0003 1 6.63 5.04 0.17 5.28 0.22 10.80 26.18 .01 .02 .68 .02 .64 .001 •C.OOOl p value 30 20 O 10 O Y////A with eggs withoul eggs Unripe Fruit with eggs without eggs Ripe Fruit (b) „ 150 £ 120 CD I 90 c 3 60 § 30 with eggs without eggs Unripe Fruit with eggs without eggs Ripe Fruit (C) 50 §- 40 < Propensity to engage in contests Contests Ripeness Contests X ripeness Infest Contests X infest Ripeness X infest Contests X ripeness X infest 30 8 20 § 10 CO existing sites. However, the pattern of attempts is illustrative. Proportionally fewer attempts were made in punctures on ripe berries than on unripe ones (Figure 3b; G = 20.17, p < .0001), as one might expect if punctures were particularly valuable to females when the intact fruit surface was difficult to penetrate. The difference in placement of attempts on ripe versus unripe berries notwithstanding, a substantial proportion of attempts were made in punctures on ripe berries (>50%; Figure 3b). This result indicates that, even on ripe fruit, flies behaved as though there was an advantage to using punctures. Finally, on both ripe and unripe fruit, proportionally more attempts made in existing punctures resulted in egg deposition than did attempts made at new sites (Figure 3c). The difference appears greater on unripe berries than on ripe ones. However, too few clutches were deposited on parasitized berries in the ripe fruit treatment to permit a statistical test of that difference. Patterns in female-female interactions with eggs without eggs Unripe Fruit with eggs without eggs Ripe Fruit Figure 2 Effects of fruit ripeness on use of egg-occupied versus unoccupied fruit (a) number of clutches laid, (b) number of egg-laying attempts, and (c) percent egg-laying attempts that result in egg deposition. punctures (Figure 3b), suggesting that the observed tendency for clutches to be deposited in punctures was due to an active preference for punctures on the part of females on unripe berries (and not, for example, to a trial-and-error process in which females repeatedly attempted egg laying in an unripe berry until they blundered into a puncture and successfully laid eggs). Finally, almost a quarter of all egg-laying attempts in punctures on unripe berries resulted in egg deposition, compared to none of the attempts made at new sites on such fruit Too few clutches were deposited on parasitized berries in the ripe fruit treatment to assess their distribution at new and Contests were generally more common on the preferred berry type within a ripeness category (Table 2). On ripe fruit, most contests took place on unparasitized berries; on unripe fruit, by contrast, most contests took place on parasitized berries. This pattern was due in part to the fact that females on the preferred fruit type simply tended more often to encounter another fly than did females on the other fruit type (Table 2). However, results also suggest that, given an encounter with another female, females tended more often to contest fruit of the preferred type than did females on fruit of the other type (Table 2). On unripe fruit, the proportion of encounters on parasitized fruit that involved a contest exceeded that on unparasitized fruit. On ripe fruit, just the opposite pattern was observed: the proportion of encounters on unparasitized fruit that involved a contest exceeded that on parasitized fruit In short, females behaved as though the more preferred substrate for egg laying within a given ripeness category was worth a relatively greater investment in terms of defense from other females. These inferences were supported by a log-linear analysis of three factors: occurrence of contest given encounter (yes versus no), infestation (parasitized versus unparasitized), and ripeness (unripe versus ripe). There was a significant inter- 240 Behavioral Ecology Vol. 7 No. 3 (a) 25 •g 20 Note: all data for occupied fruit only ® 15 o J2 action among the three factors (Table 3), indicating that the manner in which infestation influenced the occurrence of a contest depended significantly on the ripeness of the berries. There was also a significant infestation X ripeness term (Table 3), indicating that the relative frequency of female-female encounters on parasitized and unparasitized fruit depended on berry ripeness. 10 Puncture use and time saved on ripe versus unripe fruit O 6 5 0 existing new site site Unripe Fruit Ripe Fruit (b) £125 E | I 75 50 S 25 existing site new site Unripe Fruit existing site new site Ripe Fruit 25 (c) Q. 20 5 CD 10 CO a? existing site new site Unripe Fruit existing site new site Ripe Fruit Figure 3 Effects of fruit ripeness on placement of clutches at new sites versus existing egg-laying sites: (a) number of clutches laid, (b) number of egg-laying attempts, and (c) percent egg-laying attempts that result in egg deposition. Note that the data are for punctured, marked fruit only. As in thefield-cageexperiment, the tendency for females presented with punctured, marked fruit in a laboratory assay to attempt egg laying in punctures depended on fruit ripeness. On ripe fruit, only about one-third of egg-laying attempts were made directly into existing punctures; on unripe fruit, by contrast, over three-quarters of the attempts were made directly into punctures (Table 4; Gtest, G = 23.66, df = \,p< .0001). Unripe coffee berries were evidently so impenetrable that no test female successfully deposited clutches at new sites in a single attempt (Table 4). Rather, new sites were established in steps. First, a small wound was established by a female followed by repeated attempts at the same site until eggs were laid successfully. By contrast, ripe berries were penetrable enough that almost one-quarter of our test females successfully laid eggs in their first attempt (Table 4). More attempts in both ripeness treatments resulted in egg deposition in both treatments when those attempts were made at existing sites (Table 4). Because females failed to lay eggs in single attempts at new sites on unripe fruit, the time required to deposit eggs in single attempts at such sites could not be estimated. However, we obtained precise estimates for new sites on ripe fruit and in punctures on ripe and unripe fruit. There were significant differences in these measures (Table 4; one-way ANOVA, F237 = 25.44, p < .0001). Contrasts indicated that the time required to deposit a clutch into a puncture, whether on ripe or unripe fruit, was highly significantly shorter than the time required to deposit a clutch at a new site (Table 4; means comparisons by Tukey-Kramer method, p < .0001 for each comparison). Thus, even on ripe fruit there was an advantage in depositing eggs in punctures in terms of time savings. For clutches deposited in punctures, by contrast, time for deposition in ripe fruit was not significantly different from that in unripe fruit (Table 4; means comparison by Tukey-Kramer method, p > .5). DISCUSSION Functional aspects of dynamics in host acceptance Recent efforts to understand patterns of host use in insects (reviews by Godfray, 1994; van Alphen and Visser, 1990) are filling gaps in our knowledge about the dynamics of parental investment. Our own study contributes to these efforts in at least two ways. First, in most previous work, females have been Table 4 Percentage of attempts made at existing sites, percentage of attempts resulting in egg deposition at new versus existing egg-laying sites, and time required to deposit a clutch in new versus existing egg-laying sites for ripe and unripe coffee berries Attempts made at existing sites Attempts ending in egg deposition Time to oviposition (s) At new sites At existing sites At new sites In existing sites Berry type % N % N % N Median (SE) A' Median (SE) N Ripe Unripe 36 82 53 50 24 0 34 9 63 32 19 41 175 (24.4) 12 76.5 (7.8) 51.5 (6.9) 16 12 Papaj and Messing • Dynamics of superparasitism in fruit flies shown to change from a tendency to discriminate against parasitized hosts to, at best, a failure to discriminate between parasitized and unparasitized hosts. In this study, by contrast, an experimental population of medfly was observed to shift from almost absolute discrimination against parasitized hosts to almost absolute discrimination in favor of such hosts. In a sense, our results extend the range of alternative behavioral strategies known to be adopted by females of a given parasite species (see also Takasu and Hirose, 1988). Second, the noted shifts in use and avoidance of parasitized hosts were wholly consistent with a priori predictions about the costs and benefits of the alternative strategies. In other words, as has also been found in other studies of dynamics in superparasitism (Roitberg et al. 1992, 1993; van Alphen and Visser, 1990; Visser et al., 1990), shifts were functional in form. On small ripe fruit, for instance, females appear to act overtly in the interests of their young, avoiding parasitized fruit and reducing competition to those young. In this case, females avoided superparasitism under conditions in which we expected avoidance to be functional. On unripe fruit, in contrast, females appear to act overtly against those interests, using existing sites and thereby increasing competition incurred by those young. In the latter instance, females are potentially increasing the rate at which eggs are laid at the expense of the fitness of any one offspring. In this case, females superparasitized under conditions in which we expected superparasitism to be functional. Maternal versus offspring interests Egg-laying decisions in parasitic insects have been viewed as products of parent-offspring conflict (Godfray and Parker, 1992; Roitberg and Mangel, 1993; Rosenheim, 1993). As framed above, the observed shifts in female fly behavior reflect changes in the degree of conflict between parent and offspring. Critical to the notion that parent-offspring conflict in our system is dynamic is the premise that larvae of one clutch compete with members of another clutch. For example, if larvae of successive clutches cooperate in feeding within the fruit, a female's decision to place clutches in the company of other clutches would not constitute a conflict between her and her offspring. While it is not inconceivable that larvae within a clutch benefit by being in the company of other clutches (Papaj, 1993), there is substantial evidence of larval competition for the medfly (Debouzie, 1977a,b, 1978, 1981; Papaj et al. 1989a). Moreover, medfly females behave as if superparasitism is costly. Host-marking pheromone laid down on a fruit after eggs are deposited has a consistently deterrent effect even in situations where females are attracted to existing egg-laying sites (Prokopy et al., 1978; Papaj et al., 1989b, 1992). In other words, even as medfly females use existing sites, they deposit a chemical signal that reduces further use of those sites. Females may even use the level of marking pheromone on the fruit as an indication of the number of previously laid clutches and thus of the magnitude of competition likely to be incurred by their offspring (Averill and Prokopy, 1987b, 1988; Papaj, 1993). In addition, our finding that avoidance was less pronounced on large berries than small ones is exactly what would be expected if larvae compete with one another and if competition at a given egg density declines as fruit size increases. Since it would presumably always be in an individual offspring's interest to be deposited in unparasitized fruit, the fact that females adjust avoidance of parasitized fruit according to fruit size implies that some female interest (perhaps having to do with the advantages of using existing sites) is traded off against the interests of the female's offspring. Virtually all evidence for larval competition is based on stud- 241 ies with ripe fruit. It is conceivable that the restriction in use of existing sites to unripe fruit reflects a pattern in which larvae compete in ripe fruit but facilitate each other's growth and survival in unripe fruit While unripe fruit are less suitable for larval development than ripe fruit (Mourikis, 1965; Tsitsipis, 1989), there is no evidence that the pattern of fitness effects of prior clutches changes from one of competition in ripe fruit to one of facilitation in unripe fruit While the issue should be examined further, it is worth pointing out that growing evidence indicates that use of existing sites confers direct benefits upon females in terms of time savings and egg-laying success (Table 4; Papaj, 1993; Papaj et al., 1992). Surely such benefits account, at least in part, for the patterns observed here. Clutch size and patterns of egg allocation In addition to deciding where to deposit clutches, parasitic insects modify the size of clutches laid in hosts in functional ways (review by Godfray, 1987, 1994; Papaj et al. 1989b, 1990). We did not evaluate patterns of clutch size in these experiments, primarily because it is difficult to measure the size of a clutch deposited at a site already containing eggs. Owing to this constraint, we are left to assume that the patterns observed here in terms of allocation of clutches translate to similar patterns in terms of the overall allocation of eggs. This assumption seems reasonable. In medfly, clutch size (which averages about three to four) appears to be highly constrained. An earlier study (Papaj, 1990), for example, failed to find any adjustment to clutch size over a range of fruit sizes an order of magnitude greater than that examined here. Papaj et al. (1989b, 1990) did find that clutches in infested fruit were smaller than those in uninfested fruit. This effect, if it occurred in our fruit ripeness experiment, would have tended to make the difference in the total numbers of eggs laid in infested versus uninfested fruit less pronounced than the observed difference in the number of clutches laid (Figure 2a). However, since the observed preference for infested fruit in that experiment was absolute (i.e., no eggs laid at all in uninfested fruit), any effect of infestation on clutch size would not diminish the preference of females for infested fruit In short, while one might expect smaller clutches to be deposited on infested fruit, such reductions do not diminish the significance of the patterns of clutch allocation observed here. Closing remarks Our findings are consistent with correlational data reported recendy by Lalonde and Mangel (1994) who found that use of existing sites by females of another tephritid fly, RhagoUtis complela, waned over the course of a season as host walnuts ripened and the penetrability of walnut husks increased. In that study, female behavior was not observed direcdy but was inferred from the distribution of eggs in fruit Our experimental manipulations offer direct experimental support for Lalonde and Mangel's (1994) contention that use of existing sites by individual females varies as a function of fruit ripeness. It would be useful to look in turn for temporal as well as spatial variation in medfly's use of existing sites. Our experiments raise the intriguing possibility that individual females can change from avoidance of superparasitism to preference for superparasitism over relatively short periods of time. Such flexibility might be especially beneficial for a generalist like the medfly whose many host species probably vary gready in ways that affect the decision to superparasitize. We thank Henar Alonso-Pimentel, Laurie Henneman, Robert Lalonde, Cesar Nufio, Bob Smith, and an anonymous reviewer for comments on earlier drafts. 242 REFERENCES Averill AL, Prokopy RJ, 1987a. Intraspecific competition in the tephritid fruit fly, Rhagoletis pomonetia. Ecology 68:878-886. Averill AL, Prokopy RJ, 1987b. Residual activity of oviposition-deterring pheromone in Rhagoletis pomonella and female response to infested fruit. J Chem Ecol 13:167-177. Averill AL, Prokopy RJ, 1988. Factors influencing the release of hostmarking pheromone by Rhagoletis pomonella flies. J Chem Ecol 14: 95-111. Back EA, Pemberton CE, 1915. Susceptibility of citrous fruits to the attack of the Mediterranean fruit fly. J Agric Res 3:311-330. Christenson LD, Foote RH, 1960. Biology of fruit flies. Annu Rev Entomol 5:171-192. Clutton-Brock TH, 1991. The evolution of parental care. Monographs in behavior and ecology. Princeton, New Jersey: Princeton University Press. Debouzie D, 1977a. Effect of initial population size on Ceratihs productivity under limited food conditions. Ann Zool Ecol Anim 9: 367-381. Debouzie D, 1977b. Etude de la competition larvaire chez Ceratitis capilata (Diptere, Trypetidae). Arch Zool Exp Gen 118:315-334. Debouzie D, 1978. Variabilite a l'interieur d'une population de Ceratitis capitata elevee sur un milieu naturel non renouvele (Diptera: Trypetidae). Ann Zool Ecol Anim 10:515-524. Debouzie D, 1981. Analyse experimentale de l'utilisation des resources dans un systeme simphfe forme d'une banane attaquee par la mouche mediterraneenne des fruits Ceratitis capitata. ActaOecol Gen 2:371-386. Godfray HCJ, 1987. The evolution of clutch size in parasitic wasps. Am Nat 129:221-233. Godfray HCJ, 1994. Parasitoids: behavioral and evolutionary ecology. Princeton, New Jersey: Princeton University PTess. Godfray HCJ, Parker GA, 1992. Sibling competition, parent-offspring conflict and clutch size. Anim Behav 43:473-490. Hagen KS, Allen WW, Tassan RL, 1981. Mediterranean fruit fly: the worst may be yet to come. Calif Agric 35:5-7. Lalonde RG, 1992. Host selection behavior of a thistle-feeding fly: choices and consequences. Oecologia 90:534-539. Lalonde RG, Mangel M, 1994. Seasonal effects on supcrparasitism by Rhagoletis completa. J Anim Ecol 63:583-588. Liquido N, Shinoda LA, Cunningham RT, 1991. Host plants of the Mediterranean fruit fly (Diptera: Tephritidae): an annotated review. Lanham, Maryland: Entomological Society of America. Mourikis P, 1965. Data concerning the development of the immature stages of the Mediterranean fruit fly (Ceratitis capitata (Wiedemann) (Diptera: Trypetidae)) on different host fruits and on artificial media under laboratory conditions. Ann Benaki Phytopathol Inst 7:59-105. Papaj DR 1990. Fruit size and clutch size in Ceratitis capitata. Entomol Exp Appl 54:195-198. Papaj DR, 1993. Use and avoidance of occupied hosts as a dynamic process in tephritid flies. In: Insect-plant interactions, vol. 5 (Bernays EA, ed). Boca Raton, Florida: CRC Press; 25-46. Papaj, DR, 1994a. Optimizing learning and its effect on evolutionary change. In: Behavioral mechanisms in evolutionary ecology (Real LA, ed). Chicago: University of Chicago Press; 133-153. Behavioral Ecology Vol. 7 No. 3 Papaj DR, 1994b. Oviposition-site guarding by male walnut flies and its possible consequences for mating success. Behav Ecol Sociobiol 34:187-195. Papaj DR, Averill AL, Prokopy RJ, Wong TTY, 1992. Host-marking pheromone and use of previously established oviposition sites by the Mediterranean fruit fly (Diptera: Tephritidae). J Insect Behav 5:583-598. Papaj DR, HendrichsJ, Katsoyannos B, 1989b. Use of fruit wounds in oviposition by the Mediterranean fruit fly. Entomol Exp Appl 53: 203-209. Papaj DR, Roitberg BD, Opp SB, 1989a. Serial effects of host infestation on egg allocation by the Mediterranean fruit fly: a rule of thumb and its functional significance. J Anim Ecol 58:955-970. Papaj DR, Roitberg BD, Opp SB, Prokopy RJ, Aluja M, Wong TTY, 1990. Effect of marking pheromone on clutch size in the Mediterranean fruit fly. Physiol Entomol 15:463-468. Price PW, 1970. Biology and host exploitation by Pleolopkus indistinctus (Hymenoptera: Ichneumnonidae). Ann Entomol Soc Am 63: 1502-1509. Prokopy RJ, ZieglerJR, Wong TTY, 1978. Deterrence of repeated oviposition by fruit-marking pheromone in Ceratitis capitata. J Chem Ecol 4:55-63. Roitberg BD, Mangel M, 1987. On the evolutionary ecology of marking pheromones. Evol Ecol 2:289-315. Roitberg BD, Mangel M, 1993. Parent-offspring conflict and life-history consequences in herbivorous insects. Am Nat 142:443-456. Roitberg BD, Mangel M, Lalonde RG, Roitberg CA, van Alphen JJM, Vet L, 1992. Seasonal dynamic shifts in patch exploitation by parasitic wasps. Behav Ecol 3:156-165 Roitberg BD, Prokopy RJ, 1983. Host deprivation influence on response of Rhagoletis pomonella to its oviposition-deterring pheromone. Physiol Entomol 8:69-72. Roitberg BD, Prokopy RJ, 1987. Insects that mark host plants. BioScience 37:400-406. Roitberg BD, Sircom J, Roitberg CA, Mangel M, van Alphen JJM, 1993. Life expectancy and reproduction. Nature 364:108. RosenheimJA, 1993. Single-sex broods and the evolution of nonsiblicidal parasitoid wasps. Am Nat 141:90-104. SAS Institute, 1985. SAS user's guide. Cary, North Carolina: SAS Institute. Silvestri F, 1914. Report of an expedition to Africa in search of the natural enemies of fruit flies (Trypaneidae), with descriptions, observations and biological notes. Hawaii Board of Agriculture and Forestry Division of Entomology Bulletin No. 3. Sokal RR, Rohlf FJ, 1981. Biometry. San Francisco: WH Freeman. Takasu K, Hirose Y, 1988. The parasitoid Ooencyrtus nezarae (Hymenoptera; Encyrtidae) prefers hosts parasitised by conspecifics over unparasinzed hosts. Oecologia 87:319-323. Tsitsipis JA, 1989. Nutrition, requirements. In: World crop pests, vol. 3A, Fruit flies: their biology, natural enemies and control (Robinson AS and Hooper G, eds). New York: Elsevier. van Alphen JJM, Visser ME, 1990. Superparasitism as an adaptive strategy for insect parasitoids. Annu Rev Entomol 35:59-79. Visser M, ran Alphen JJM, Nell HW, 1990. Adaptive superparasitism and patch time allocation in solitary parasitoids: the influence of the number of parasitoids depleting a patch. Behaviour 114:21-36.
© Copyright 2025 Paperzz