PATCH TIME ALLOCATION AND SEARCH ... - Frietson Galis

PATCH TIME INTENSITY

ALLOCATION OF ASOBARA A LARVAL (BRACONIDEA), OF DROSOPHILA

AND SEARCH TABIDA NEES PARASITOID

by F. GALIS and J. J. M. VAN ALPHEN (Departmentof Ecology,ZoologicalLaboratory,Universityof Leiden, The Netherlands SUMMARY We studied three factors affecting the allocation of patch time and the intensity of searching by the larval parasitoid Asobaratabida Nees. We found that: 1.A. tabida reacts to a water-soluble kairomone, produced by its host DrosophilamelanogasterMeigen in that larvae had crawled and a) time spent on a patch (yeast patch in which D. melanogaster fed) increased in a S-shaped fashion with increasing kairomone concentration. b) the searching intensity by the parasitoid increased with increasing kairomone concentration until it levels off at higher concentrations. 2. A. tabida recognized areas previously searched by a conspecific and spent less time and searched less intensely on such patches as compared to unsearched patches. 3. Patches with a reduced quality (reduced amount of living yeast) for the host are less attractive for the parasitoid which spends less time and searches less intensely on such patches than on patches with a better quality for the host. The response of the parasitoid to these three factors contributes to the optimization of time allocation. The increase in search intensity can eventually cause a sigmoid functional response and may enhance population stability. INTRODUCTION

Currently much interest is focused on the allocation of searching time or predators. Most research is between patches by insect parasitoids concerned with the question, what is the best strategy for the parasitoid effort (e.g. ROYAMA, 1970; HASSELL from the standpoint of reproductive & MAY, 1974; MURDOCH & OATEN, 1975; CHARNOV, 1976; COOK & HUBBARD, 1977; VAN LENTEREN & BAKKER, 1978; COMINS & HASSELL, 1979 ; WAAGE, 1979; TINBERGEN, 1980 ; VAN ALPHEN, 1980), whilst other research is concerned with the question, what is the influence on the dynamics ofparasitoid-host interaction (e.g. HASSELL &MAY, 1974; MURDOCH & OATEN, 1975; COMINS & HASSELL, 1979). It is generally believed that for parasitoids to spend more time in denser patches than in sparse ones in functional and that this behaviour promotes the stability of parasitoid-host dynamics. to make two Allocating time on a patch may force the parasitoid

597 decisions (KREBS et al., 1974): firstly it should decide which patch to visit; secondly it should decide when to leave a patch. The first decision is only feasible if the parasitoid can predict the patch quality by cues which can be perceived from a distance. Host-associated cues are known to play an important role in attractingl to parasitoids patches and in to arrestl on a patch (e.g. VINSON, 1976; causing the parasitoids of host-associated cues varies with WAAGE, 1978). If the concentration host density and the parasitoid shows a concentration-dependent response to these cues, then this may provide one of the behavioural mechanisms used to allocate patch time (see WAAGE, 1978, 1979). In addition to causing the parasitoid to become arrested, chemical cues are known to play a role in distinguishing previously searched from unsearched areas. For example PRICE (1970) found that Pleolophus basibetween previously searched and unsearched zonus discriminates host habitat areas, in that it avoids presearched ones. This behaviour is based on the recognition of a mark left during a prior visit by itself or a conspecific wasp. A similar behaviour has been found for the parasitoid Diaparsis truncatus (vart ALPHEN, 1980). Asobara tabida Nees, a solitary larval parasitoid of Drosophila species is attracted to and arrested by fermenting fruit, the medium for its host. However, once arrested the parasitoid must determine whether potential hosts are present in the patch. One potential mechanism which In might facilitate this is the response of the parasitoid to kairomones. the present study we wished to determine if a kairomone existed and if so, whether the amount of kairomone or the response to it by the parasitoid was related to the density of host larvae that had been present in the patch. Furthermore we also wished to determine whether a prior visit to the patch by a conspecific effected the allocation of searching time to a patch.This the foraging paper is one of a series investigating of A. its to host tabida, including strategy response density, to different host species and to a patchily distributed host. HOST SEARCHING

BEHAVIOUR

OF A. tabida WITHIN

A PATCH

When a female A. tabida steps onto a patch of yeast she changes her manner of locomotion: she stops every few millimeters. When the yeast turns and spot has been crossed and its edge reached, the parasitoid crosses the patch in a different direction. Then host location proceeds as 1 Attraction in the sense of causing the parasitoid to make oriented movements towards the patch and causing to arrests in the sense of slowing the linear progression of the parasitoid by reducing actual speed of locomotion and/or by increasing rate of turning (DETHIERet al., 1960).

598 follows: when the wasp is stationary she perceives moving host larvae from a distance and reacts by turning and walking towards the moving host. This behaviour involves no visual stimuli, hence A. tabida is capable of detecting hosts which are hidden under a layer of yeast, or in yeast-filled crevices in the agar. The positive relation between the as the number of jaw-movements per activity of a larva (measured indicate minute) and the chance of being detected by the parasitoid that A. tabida reacts to vibrations in the host's medium caused by the larvae (VAN ALPHEN et al., in prep.). Sometimes a host larva is located while the wasp is walking, i.e. when she steps on one. In this case slight movements of the host, a kairomone associated with it and/or textural cues from the host's skins are all possible stimuli to which the parasitoid might react. When a host is located, A. tabida probes for it with the a ovipositor until it pierces the host or, if unsuccessfull in contacting host, gives up. Often a series of stabs with the ovipositor is necessary Such probing bouts may alternate with before the host is contacted. host location behaviour. For a more detailed description of host searching behaviour of A. tabida see van ALPHEN et al. (in prep.).

MATERIAL

AND TECHNIQUES

Adult female A. tabida wasps, Leiden strain (for origin and rearing, VAN ALPHEN, in prep.) were stored at 10°C until required for experimenthe wasps were allowed to tation. One day prior to an experiment in h. at to hosts for 1.5-2 20°C gain experience (see SAMSONoviposit females were then kept at 20°C BOSHUIZEN et al., 1974). Experienced and moisture. Early second with sugar waters source of carbohydrates instar larvae of D. melanogaster, strain WW were used as hosts (for rearing, see BAKKER, 1961). The first series was deWe carried out three series of experiments. signed to determine whether a parasitoid could detect a patch in which larvae had been present, and whether the time the wasps spent on the patch was related to the density of larvae that had occupied it. To Series (a): a this we designed the following experiments. determine viscous suspension of yeast, 2 cm. diam., was placed on a layer of agar, the latter contained within a 5 cm. diam. petri-dish. The yeast suspenor sion contained 0.125 gr. of yeast. In these patches 0, 1, 2, 4, 8, 16, 32 64 host larvae, enclosed in perspex rings (2 mm high), were allowed to crawl and feed. After 18 h. the larvae and perspex rings were removed conand the patch was offered to individual wasps. The experiments sisted of exposing a patch to a single female A. tabida. They were terminated when a wasp left the patch for more than a minute or if she

599 when a parasitoid to migrate. This latter was manifested attempted walked to the lid of the petri-dish and attempted to leave. In practiWe reboth the above criteria were applicable. cally all experiments corded the experiments on videotape while simultaneously observing the parasitoid's behaviour with the aid of a stereo microscope. Afterwards the paths of the wasp, displayed on video monitor, were traced were behavioural on a plastic sheet. The following components measured: (2) time on the patch, (3) time (1) duration of experiment, walked (5) distance spent motionless, (4) time spent walking, (6) walking speed ( (5) / (4) ), (7) (measured with a map measurer), number of stops, (8) average duration of a stop, (9) average duration of interval between stops, (10) average distance between stops, (11) number of probes, (12) average number of degrees turned per 0.5 cm. (by measuring the angles between tangents drawn at 0.5 cm intervals along the path). Series (b) : in part to explain the results of series (a) and to determine whether the host associated cue used by the parasitoid was physical or chemical a similar set of experiments to series (a) was carried out at larval densities of 0, 2, 4, 32 and 64. However, after the larvae and perspex ring had been removed, 1.5 ml of water was added to the yeast and this fluid was filtered. To the filtrate, we added an amount of fresh yeast of the same age, equal to the amount removed during filtering. The excess water was evaporated to provide a suitable substrate for the We choose this procedure to make the experisearching parasitoid. ments comparable to series (a). However, by this procedure the total amount of yeast cells in the patches decreased with the increasing density of larvae that had been in the patch. This could have had a detrimental effect on patch time at the patches which had contained a inhigh density of host larvae, but other (unpublished) experiments dicate that the effect is negligable. We measured total time spent on the patches that had contained 2 and 32 larvae. On the patches that had contained other densities all 12 variables were measured. The patches containing no larvae (density 0) were also filtered to check whether behaviour. filtering itself had any effect on the parasitoid's Again 10 were carried at each of the 5 larval densities. out replicates Series (c) : in addition a third series of experiments was conducted. It was designed to determine whether a prior visit to the patch by a conspecific effected the response of the parasitoid to the patch. Patches expreviously visited by a conspecific (series (a)) were subsequently posed to a second female wasp. The patches used in the experiment had contained 4 larvae prior to their exposure to the first female wasp.

600 RESULTS

AND DISCUSSION

Time on the patch A. tabida exhibited an exponential increase in the average time spent on the patch with the increasing number of larvae that that patch had originally contained (fig. 1; table 1). This increase levelled off at those patches that had contained 4 or more larvae. We conclude from these results that the wasps responded to cues present in the yeast patches. These cues may be stimuli of different origins, all associated with the host larvae (e.g. exuviae, dead yeast excreted by feeding larvae or metabolic waste products). The results show an unexpected decrease in time at host densities. To this and to elucidate the patch high explain nature of the stimulus to which A. tabida responds, the experiments with filtered yeast (series (b)) were done. We observed a similar pattern in the responses of wasps to the filtered as well as the unfiltered yeast (fig. 1, table 1). The only significant difference between these occurred with 64 larvae (fig. 1): wasps exposed to unpatches that had contained

Fig. I A. The average time on patch is shown against the larval densities at which the kairomone was produced. I.B. The lines indicate the larval densities at which the wasps did not differ significantly for the time on patch (a = 0.05). filtered yeast patches remained on these a shorter period of time than on filtered ones. We conclude from this similarity that the material to which the parasitoid responded is a watersoluble kairomone. We interpret the difference between the parsitoid's response at the patch with kairomone derived from 64 larvae, as one that is due to a response by the parasitoid to the different amounts of live yeast present in the un'filtered and filtered patches. Because the response to the kairomone increased after replacing dead or excreted yeast cells by fresh ones this

601

602

603 suggests that the parasitoid responded to both a kairomone and a cue associated with the amount of larval food available. That this response to available food is functional can be seen from the results of related (VAN ALPHEN & GALis, in prep.) in which we have shown experiments that D. melanogaster larvae emigrate from exploited areas (i.e. ones in which the yeast has been extensively fed on) to adjacent areas containing fresh yeast. Hence, areas composed of mainly dead and used yeast are unlikely to contain suitable host larvae and due to this not worth allocating search time. The mechanism that determines patch time may be a waning response to the patch edge, where the wasp turns inward when the patch edge is detected. Fig. 9 shows the walking paths resulting from this response. WAAGE (1978) described a similar response for Nemeritis canescens. The variation in the response of individual wasps increased with the larval density at which the kairomone was produced (see sd in table 1). This is in contrast to the variation in patch time in a series of unpublished experiments in which kairomone as well as host larvae were present. We therefore believe the variation in the experiments discussed here to be caused by the unnatural circumstance that kairomone occurs in the absence of host larvae. (Though the distribution is not normal we calculated the standard deviation as a measure to express the variation.) Time spent standing still and walking The time spent on the patch is divided

into that spent walking

and that

Fig. 2 A. The average percentage of time spent standing still is shown against the larval densities at which the kairomone was produced. 2.B. The lines indicate the larval densities at which the wasps did not differ significantly for the percentage of time spent standing still (a = 0.05).

604

Fig. 3.A. The average time spent standing still is shown against the larval densities at which the kairomone was produced. 3.B. The lines indicate the larval densities at which the wasps did not differ significantly for the time spent standing still (a = 0.05). spent standing still. The majority of the time spent on the patch by the wasp was spent standing still (fig. 2 and table 1). Therefore its pattern of increase was the same as that for the total patch time (fig. 3). But the proportion of the total time spent standing still increased with the larval (fig. 2). density that the patch had contained Since A. tabida while motionless, locates its hosts by sensing their movements in the medium, the amount of time spent standing still is a measure of search intensity. Thus not only does the parasitoid respond to an increase in kairomone by spending more time on a patch but also a greater portion of this time is allocated to standing still.

Fig. 4.A. The average distance walked is shown against the larval densities which the kairomone was produced. 4.B. The lines indicate the larval densities at which the wasps did not differ significantly for the distance walked (a = 0.05).

605

Fig. 5.A. The average number of stops is shown against the larval densities at which the kairomone was produced. 5.B. The lines indicate the larval densities at wh?h the wasps did not differ significantly for the number of stops (a = 0.05). A similar difference unfiltered yeast patches fig. 3 and 2).

in the response of parasitoids to filtered and was observed in this case also (compare fig. 1 vs.

Distance walked, number of stops and average duration of a stop Distance walked (fig. 4), the total number of stops (fig. 5) and the average duration of a stop (fig. 6) all showed a pattern similar to that observed for total time on a patch (fig. 1). The apparent peak observed in the average duration of a stop is not significant (fig. 6). The dip

Fig. 6.A. The average duration of a stop is shown against the larval densities at which the kairomone was produced. 6.B. The lines indicate the larval densities at which the wasps did no4 differ significantly for the duration of a stop (a = 0.05).

606

Fig. 7.A. The average velocity is shown against the larval densities at which the kairomone was produced. 8.B. The lines indicate the larval densities at which the wasps did not differ significantly for the velocity (a = 0.05). observed provide

in the average distance walked is significant no sufficient explanation.

Duration of interval and distance between stops No apparent relation was found between an increase the distance and time between stops (table 1).

fig. 4), but we can

in kairomone

and

Walking speed and average turning per distance Although walking speed differed significantly in some cases, there is no clear relation between this variable and the number of larvae that crawled through the yeast patch (fig. 7). Furthermore no significant relationship existed between the number of larvae that the patches had contained and the number of degrees turned per 0.5 cm (table 1). Number of probes The average number of probes by a parasitoid on a patch increased on the patches that had contained an increasing number of larvae (fig. 8). However, not all differences are significant, this is probably due to the fact that many wasps did not probe at all (table 1). Again with unfiltered patches that had contained 64 larvae a significant decrease in the response was noted (fig. 8). Filtering these patch types eliminated this significant decrease (fig. 8). We conclude from the observation that A. tabida probes with the in from which larvae and exuviae are removed that ovipositor patches is not elicited host movements or tactical cues from the probing only by

607

Fig. 8.A. The average number of probings is shown against the larval densities at which the kairomone was produced. 9.B. The lines indicate the larval densities at which the wasps did not differ significantly for the number or probings (a = 0.05). host's skin, but also in response to kairomone. The increasing tendency to probe with the ovipositor with increasing number of larvae that had been present in the patch (table 1) indicate that the threshold to probe is lowered in response to an increasing amount of kairomone. These changes in the threshold to probe may increase the searching e?ciency: less time is spent in probing when the probability to hit a host is low, while at kairomone concentrations associated with high host densities more time is devoted to probing. The effect of the kairomone on the behaviour of A. tabida Our conclusion that the parasitoid responds to a kairomone (see page of the parasitoids be600) is further bolstered by several observations haviour. Firstly the parasitoid spent an increasing proportion of its time standing still (fig. 2), a behaviour associated with detecting the movement of its host and secondly the wasp probed in patches from which larvae and their exuviae had been removed (by filtering) (fig. 8). Both these responses showed a pattern of increase in association with increasing larval density (i.e. larvae that had been present in the patch). This implies that the kairomone present is related to host density (at least over a range of 0 to 4 hosts per patch). That the response remains on yeast that was filtered with water implies that the kairomone is watersoluble. Moreover, the decreased response observed for the variables in unfiltered yeast patches that had contained 64 larvae was counteracted by the filtering and replacing of the used yeast with living yeast cells. This suggests that the parasitoids also can detect a factor associated with the quality of the food for its host. These patches consisting of used

608

Fig. 9.A. The paths of individual wasps are shown: A. at larval density 0; B. at larval density 4, of the first wasp that visited the patch; C. at larval density 4, of the second wasp that visited the patch. The dots indicate the position of the middle of the abdomen when the wasp stood still. Other experiments showed similar results. yeast are unlikely to contain suitable hosts and therefore rejected after detection. Therefore two cue types are integrated. The increase in probing rate and the allocation of a greater proportion of patch time to standing still both show that A. tabida searches the patch more intensively in response to higher kairomone concentrations. Searching more intensively in response to a cue associated with host increase in encounter rate at indensity may produce a proportional host densities. HASSELL et al., creasing (1977), and MURDOCH & OATEN have such a shown that ( 1975) change in encounter rate can generate a sigmoid functional response.

TABLE2 A comparison between the first and the second wasp that visited the patch for time spent on the patch and the number of probings.

609 Second visits to the patch A wasp who visits a patch, previously visited by a conspecific, stays shorther on the patch than one who visits a patch not previously visited (fig. 9. B.C.). Evidently the first leaves a mark that is recognized by the second wasp. The behaviour of the second wasp to stay short on the is functional since the patch expectation of the number of unparasitized host larvae in the patch is lower if the patch has been visited before. Like series (a) and (b) a shorter time on the patch coincides with a shorter time spent standing still and walking, a lower proportion of the time on the patch spent standing still, a shorter distance walked, a smaller number of stops and a longer average duration of a stop. Thus again the tendency to stay shorter on the patch coincides with a decrease in search intensity. There are no significant differences in walking speed, average duration of the interval and average distance between for which no clear relation with the stops, precisely those parameters amount of kairomone in series (a) and (b) could be demonstrated. the difference in the number of probings is nog significant. Unexpectedly It is notable that all patches where the first wasp had probed the second wasp also probed. There is only one patch where the first wasp dit not in probing beprobe and the second did (nr. 8). This correspondence tween the first and second wasp might be due to the same vertical position of the empty larval skins in the yeast patch. Apart from this similarity in the responses of the wasps to the same between the time on patches thee is a significant positive correlation the patch of the first wasp and of the second wasp (r = 0.79, a = 0.05, rank correlation, table 2). Possibly the wasps react to difSpearman ferences in the quality of the patch, though it is not clear what these differences are. Small differences in the amount of kairomone might be responsible but it must be kept in mind that in series (a) and (b) there is only an increase in time on patch up to larval density 4 and no significant difference between density 4 and 8. CONCLUSIONS The result of series (a), (b), and (c) lead to the conclusion that if the of the number of unparasitized host larvae is higher, then expectation the tendency to stay on the patch increases up to a maximum as does the intensity of the search. These changes in behaviour are the result of the wasp's response to the amount of kairomone, to a cue associated with the qu'ality of the medium for the host (amount of available food) and to the presence or absence of a mark of another wasp that visited the patch previously. This means that even without actually encounterabout the number of hosts ing hosts the wasps can gather information

610 to quickly of time it is important present. If there is optimalization assess the profitability of the patch (KREBS et al., 1974). The timing of encounters with hosts on which threshold rate models are based (see ROYAMA, 1970; HASSELL & MAY 1974; MURDOCH & OATEN, 1975; CHARNOV, 1976) is a stochastic process which will often lead to such as, behaviour (OATEN, 1977). Additional information suboptimal the quality of the yeast and the for A. tabida, the amount of kairomone, presence or absence of a mark of another wasp diminishes the stochasticity (see also WAAGE, 1979) and diminishes the time necessary for the assessment. The effect of the timing of host encounters will be discussed later (VANALPHEN &GALrs, in prep.). The behaviour ofA. tabz*daindicates There are several reasons why time that there is time optimalization. seems functional for A. tabida: a) the high numbers of optimalization wasps searching in each baited trap exposed in the field at different places near Leiden during the summer of 1980 indicates that intraspecific competition for host larvae may be severe, b) wasps may die before all eggs are laid, c) the resources of the host larvae e.g. rotting fruit, occur and have a short existence, d) A. tabida is a preovigenic unpredictably species. host According to HASSELL & MAY (1974) the stability of parasitoid increases if there is a large difference between the maxiinteractions mum (high host density) and minimum (low host density) time spent to MURDOCH & OATEN according searching per unit area. Further, (1975) and HASSELL et al., (1977) an increase in search intensity (attack rate) may lead to a type 3 functional response (sensu HOLLING 1959) which they think may contribute to stability. The behaviour of A. tabida to stay longer on a patch and to search more intensely when a high number of host larvae is likely may thus favourably influence the stawith the hosts. bility of the interactions

ACKNOWLEDGEMENTS S

We are grateful to Dr. R. F. Luck (Riverside, U.S.A.) for his meticulous improved the article. We thank Prof. Dr. editing, which substantially K. Bakker and Dr. G. J. de Bruijn for stimulating discussions. Prof. Dr. K. Bakker, Dr. G. J. de Bruijn, Dr. Melanie Stiassny and Dr. C. D. N. for their valuable comment on previous drafts Barel are acknowledged we thank Mr. C. Elzinga, Mr. G. P. G. Furthermore of the manuscript. Hock and Mr. P. C. G. Glas for making the drawings, Mr. G. J. F. Boskamp and Dr. Willy van Strien for rearing the parasitoids and host larvae and Mrs. Geerda Beurs for the typing of the manuscript.

611 REFERENCES ALPHEN, J. J.M. VAN,1980. Aspects of the foraging behaviour of Tetrastichusasparagi Crawford and Tetrastichusspec. (Eulophidae), gregarious egg parasitoids of the L. (Chrysomelidae). asparagus beetles Criocerisasparagi L. and C. duodecimpunctata I. Host-species selection, host-stage selection and host discrimination. - Neth. J. Zool.30: 307-325. K. 1961. An analysis of factors that determine success in competition for food BAKKER, among larvae of Drosophilamelanogaster. -Arch. néerl. Zool. 14: 200-281. E. L., 1976. Optimal foraging: the marginal value theorem. - Theor. Popul. CHARNOV, Biol. 9: 129-136. H. N. & M. P. HASSELL, 1979. The dynamics of optimally foraging predators COMINS, and parasitoids. - J. Anim. Ecol. 48: 335-351. 1977. Adaptive searching strategies in insect parasites. COOK,R. M. &S. F. HUBBARD, J. Anim. Ecol. 46: 115-125. BROWNE &C. N. SMITH,1960. The designation of chemicals DETHIER,V. G., L. BARTON in terms of the responses they elicit from insects. -J. Econ. Entomol. 53: 134-136. M. P., J. H. LAWTON 1977. Sigmoid functional responses HASSELL, & J.R. BEDDINGTON, in invertebrate predators and parasitoids. - J. Anim. Ecol. 46: 249-262. M. P. &R. M. MAY, 1974. Aggregation of predators and insect parasites and HASSELL, its effect on stability. - J. Anim. Ecol. 43: 567-594. HULLING,C. S., 1959. The components of predation as revealed by a study of smallmammal predation of the European Pine Sawfly. - Can. Entom. 91: 293-320. 1974. Hunting by expectation or optimal KREBS,, J.R., J. C. RYAN&E. L. CHARNOV, foraging? A study of patch use by Chickadees. - Anim. Behav. 22: 953-964. LENTEREN, J.C. VAN& K. BAKKER,1978. Behavioural aspects of the functional responses of a parasite (PseudeucoilabocheiWeld) to its host (Drosophilamelanogaster). -Neth. J. Zool. 28: 213-233. W. W. &A. OATEN,1975. Predation and population stability. - Adv. Ecol. MURDOCH, Res. 9: 1-25. OATEN,A., 1977. Optimal foraging in patches: a case for stochasticity. - Theor. Pop. Biol. 12: 263-285. PRICE,P. W., 1970. Trail odors: recognition by insect parasites on cocoons.-Science 170: 546-547. ROYAMA, T., 1970. Evolutionary significance of predators' response to local differences in prey density: a theoretical study.-Proc. Adv. Study Inst., Dynamics Numbers. Popul. (Oosterbeek) : 344-357. &K. BAKKER,1974. Success of paeasitizSAMSON-BOSHUIZEN,, M., J. C. VANLENTEREN ation of Pseudeucoila bocheiWeld (Hym., Cynip.) : a matter of experience. - Neth. J. Zool. 24: 67-85. TINBERGEN, M., J. 1980. Foraging decisions in starlings (SturnusvulgarisL.). -Ardea 69: 1-67. VINSON,S. B., 1976. Host selection by insect parasitoids. - Ann. Rev. Entomol. 21: 109-133. WAAGE,J. K., 1978. Arrestment responses of the parasitoid, Nemeritiscanescens,to a contact chemical produced by its host, Plodia interpuctella. - Physiol. Ent. 3: 135-146. WAAGE, J.K., 1979. Foraging for patchily-distributed hosts by the parasitoid Nemeritis canescens. J.- Anim. Ecol. 48: 353-371. Authors' address: Department of Ecology, Zoological Laboratory, University of Leiden, Kaiserstraat 63, 2300 RA Leiden, The Netherlands.