HOST-FINDING BY INSECT PARASITES

Transcription

1 VOL. XV, No. JULY, HOST-FINDING BY INSECT PARASITES II. THE CHANCE OF TRICHOGRAMMA EVANESCENS FINDING ITS HOSTS BY J. LAING Research Fellow of Newnham College From the Zoological Laboratory, Cambridge {Received June ) (With Five Text-figures) INTRODUCTION IN a previous paper (Laing, ) an account was given of the way in which the chalcid parasitoid Trichogramma evanescensfinds its hosts. It was shown that the parasite is assisted in its search for hosts (eggs of moths) by its ability to see the eggs and to perceive traces left by adult female moths. It was further demonstrated that a Trichogramma leaves an egg which it has just parasitized or examined on a twisted track, winding around and away from the egg; and that this "turning movement", which occurs after an oviposition or an examination but not after a mere touch, increases the parasite's chance of finding a neighbouring host. The above data concerning the process of host-finding by Trichogramma are of interest in the field of insect behaviour, but are of little value in connexion with the interaction of host and parasite populations unless supplemented by the results of quantitative investigations. This study is an attempt to investigate the process of host-finding quantitatively, by determining the effect of variation in the size of hosts and the distance between them upon their chance of being found. MATERIAL AND METHODS The parasite used throughout the experiments is the chalcid parasitoid Trichogramma evanescens. The adult female of this species attacks the eggs of many insects, especially moths, laying an egg within them. The Trichogramma egg hatches into a larva which devours the entire contents of the host egg and then pupates within the empty egg-shell, emerging as an adult wasp about days (at C.) after the parasitic attack. All the individuals used were from the pure strain maintained by Dr G. Salt, and were reared and prepared for experiment according to the technique described by him (Salt, a). 'Throughout the experiments the discovery of Salt (), that Trichogramma females can distinguish tracks of other females, was taken into consideration, and only clean glassware and graph paper were used. Except where otherwise stated, the hosts used in this investigation were eggs of the grain moth, Sitotroga cerealella. ieb-xviii

2

3 Host-finding by Insect Parasites THE EFFECT UPON FINDING OF VARIATION IN THE DISTANCE BETWEEN THE HOSTS To determine the effect upon finding of varying the distance between the hosts, a series of experiments was carried out, using a technique already described (Laing, ). For each experiment, eggs of Sitotroga cerealella were pasted on graph paper in two rows, - in. apart, the eggs in each row being regularly spaced at either -, o-io, -, -, - or - in. apart. A female Trichogramma was placed upon the graph paper and was covered with a Petri dish, and her movements were noted and timed until she had made contacts with the eggsa contact being either an oviposition, an examination, or a mere touch. Ten experiments were performed with eggs at each of the six spacings, in daylight and at a temperature of - C. The results of the experiments are given in detail in Table I and in summary in Table II. Table II. Summary of results of Table I Distance between eggs (in.) - o-io O-O - - % of neighbouring contacts % of neighbouring contacts following an oviposition or examination % of neighbouring contacts following a touch i-o - 'S Before the results can be judged it is necessary to understand the meaning of the term "neighbouring contact" (NC), since the frequency of neighbouring contact is the chief criterion by which the effect of varying the distance between the hosts is estimated. Throughout the experiments it was particularly noted when a Trichogramma moved from any egg to a neighbouring egg directly, that is, without moving beyond the limits of a circle whose centre was the egg that was being left and whose radius was equal to the distance between the eggs. A contact made after a direct movement of this sort was noted as a "neighbouring contact". For instance, if a parasite moved between eggs,,,, and ' (Fig. ), movements to and to would be "neighbouring contacts", but movements to, to, and to ' would not. The frequency of neighbouring contact made by the Trichogramma when leaving an egg after all kinds of contactoviposition, examination and touchis measured by the value NC/C%, which is the percentage of neighbouring contacts (NC) out of the total number of contacts (C=$). It has already been mentioned (p. ) that after an oviposition or an examination, but not after a mere touch, the parasite leaves the egg on a much-twisted track, thereby increasing its chance of coming into contact with a neighbouring egg. Because of this turning movement it is of use in the analysis of the results to consider not only the value NC/C% but also the values NCJC^/o, which is the frequency of neighbouring -

4 J. LAING ' ' ' ' ' ' II' ' ' ' ' Fig.. Arrangement of Sitotroga egga in first experiment. r- - S c f o i- -" -" -" -" -" Distance between eggs -" Fig.. The effect on the frequency of neighbouring contact of variation in the distance between the eggs. O O NCJCQ.%, NC/C%, x-.-x NCJC.%, - calculated chance of NC.

5 Host-finding by Insect Parasites contact following oviposition and examination, and NC t jc t %, the frequency after touch. In Fig. the.three values NC/C%, NCJC M % and NC t /C t % are plotted against the distance between the eggs for the experiments. The main conclusion to be drawn from the result is that the frequency with which the parasite makes a neighbouring contact varies inversely with the distance between the eggs. Observation of the results shows what is the relation between distance apart and frequency of contact, but further investigation is necessary to show how this relation is brought about. It is obvious that the chance of making a neighbouring contact must decrease as the distance between the" eggs increases. The first step in the investigation therefore is to determine whether the observed change of contact frequency is entirely a result of chance. This determination has three stages, of which the first is the calculation of the distance of perception for the conditions holding during the experiments. It has been shown previously that Trichogramma perceives the eggs from a distance by sight, and it is known that the distance from which it sees them depends upon the light intensity. Since one cannot estimate the distance accurately merely by observation it is necessary to calculate its value separately for each set of experiments. When this value is found it becomes possible to proceed to the second stage, which is the calculation of the frequencies of contact which would result from random movement. The third stage is comparison of the calculated with the observed frequencies. () Calculation of the distance of perception In order to make this calculation, it is necessary to assume that a particular frequency of neighbouring contact results from random movement of the Trichogramma. The value to be used as the basis of calculation must, therefore, be a Fig. - frequency of neighbouring contact following touch, and must be obtained from the experiments in which the distance between the eggs is neither so small that it is less than the radius of perception of the Trichogramma for a Sitotroga egg, nor so large as to allow the frequency to be affected by the kind of turning which may occur in the movement from an egg after this has been merely touched. For these reasons,

6 J. LAING the frequency NC t /C t % obtained when the eggs are - in. apart is taken to be a value resulting from random movement. Then, if A (Fig. ) is the centre of the egg from which a Trichogramma is moving, B is the centre of one of the neighbouring eggs, and BC the radius of perception of the insect for the egg at B, the chance of a contact with this one neighbouring egg, B, ABAC is. When AB = o-is in., L f AC = *' (Table II). Therefore D x v ' i; x ' and BC (radius of perception) = sin L - x - in. = - in. This figure is of the order one would expect from observation. () Calculation of the chances of contact If the parasite can see an egg from a distance of - in. from either side, the chance of a contact with either of the two neighbouring eggs should be % when the eggs are arranged in one line at - in. apart. x Z. BAC The chance of a neighbouring contact on either side is ^- x % and may be calculated for each of the distances o-i to - in. as follows: When AB is o-i in., sin L BAC = ^^; and L ^ =. Therefore the chance is v - x % = -%. Similarly, when AB = o-in., the chance is -%; when AB = o- in., it is -%; and when AB = o-^ in., it is -%. The values calculated for the chance of contactare plotted in Fig.. () Comparison of the frequencies obtained with the chances calculated There is a general likeness between the curve of the frequency of neighbouring contact obtained by experiment and that showing the frequency to be expected from chance movements (Fig. ). The two curves are sufficiently alike to warrant the conclusion that the relation between the observed frequency and the distance between the eggs is very largely governed by chance. There are, however, two discrepancies between the absolute frequencies observed and the chances expected: () when the eggs are - in. apart, the observed value is less than that calculated; () when they are o- in. or further apart, it is greater. () Examination of the data of the experiments with eggs at - in. apart shows that, even when the departures from the end eggs of the rows are eliminated, the movements from eggs each possessing two neighbours do not always result in

7 Host-finding by Insect Parasites a neighbouring contact. Thirty contacts that are not neighbouring contacts were made from eggs that were not end eggs. Of these, eighteen followed an oviposition or examination, and twelve followed a touch. Although it is not actually known whether in these cases the parasite did not see, or whether it saw but did not move towards the neighbouring eggs, the latter alternative appears the more probable. That is, it is likely that perception occurred but no reaction followedthe Trichogramma did not make use of its faculty of vision. In any case, it is clear that the parasite does not invariably move from one egg directly to another within the range of its perception, and hence does not automatically restrict its searching to one group of eggs. () The calculated value for frequency of neighbouring contact is the probability that a Trichogramma moving in a straight path from one egg will meet either of the neighbouring eggs, whereas the observed value includes the contacts made by the parasite moving in the twisted path which follows an examination or oviposition. When the eggs are very close (- in.), the parasite moves direct to the next egg more frequently after a touch than after an oviposition; but at all other spacings the turning movement increases the frequency of neighbouring contact. On account of the turning movement, therefore, at spacings of o- in. or more the observed frequency is greater than the calculated. The extent to which the frequency of contact is increased by the turning movement depends on the distance between the eggs. The proportion of neighbouring contacts following an oviposition or examination to those following a touch varies with the distance as follows: Eggs o-i in. apart, NC^/C^ : NC t /C t = -: - =-: - =-: - =i-:.. - =: The maximum effect of the turning movement occurs when the eggs are - in. apart, at which distance it doubles the chance of a neighbouring contact. When the eggs are closer together, the chance is greater that the parasite will find the second host before it has completed its turning movement, and the manoeuvre is wasted. When they are farther apart, the proportion decreases a little, chiefly by reason of the high values of NC t jc t which are produced at the greater spacings by the ordinary deviations of the parasite from a straight line. In summary, the experiments of this section have shown that the general relation between the frequency of neighbouring contact and the distance between the eggs is governed chiefly by chance; and that the differences between the frequencies observed and those calculated on the basis of random movement result from the peculiar nature of the movements of the parasite.

8 J. LAING THE RATE OF FINDING To this point the effect on finding of variation in the distance between the hosts has been assessed by the proportion of neighbouring contacts to the total number of contacts. This ratio was chosen as criterion because it made possible an analysis of the means by which the factor of distance exerted its effect. It is, however, not an absolute measure, and now, in order to bring the results of the previous experiments to an absolute scale, it is necessary to introduce the time factor and to consider the rate of finding hosts. The rate at which the parasite finds hosts depends upon two factors: (i) how quickly it moves, and () how far it has to move from one host to another. Although Trichogramma has wings it does not use them for movement from host to host, and the rate to be measured, therefore, is that of walking. The first factor, the rate of movement of Trichogramma, was measured in two series of experiments in daylight and at a temperature between and C. The first series was carried out under the same conditions as those of the experiments described in the foregoing section. A female Trichogramma was allowed to move about under a Petri-dish cover, on graph paper on which were pasted from to eggs.,as she walked to and from the eggs, the track of her movements was drawn as closely as possible on another similar piece of graph paper, while the time of each separate journey was noted with a stop-watch. Ten parasites were timed, each for about min. of movement. The curved tracks which had been drawn were then measured by means of fine dividers. The rates of movement for the parasites were, in inches per minute, -, -, -o, -, -, -, -, -, io- and II-I respectively, the average rate being - in. per min. or - in. per sec. The experiments of the second series were carried out as follows: isolated female parasites were confined within Petri dishes and these placed inside an incubator with a double glass top. The average temperature inside the incubator was C, with a variation of not more than o- C. The top of each Petri dish was marked in -^ in. squares and the time taken by a parasite to walk, straight, a distance of in. was measured with the aid of a stop-watch. Five such measurements were obtained for each of eight parasites. The rate of movement for each of the insects was, in units of tenths of an inch per second, -, -, -, -, *, -, - and -; the average rate being - units or - in. per sec. Both series of experiments gave exactly the same result of - in. per sec, and this value will be used in subsequent calculations. The second factor, how far the parasite has to move from one host to another, depends upon the distance between the hosts. Experiments already described on p. serve to show the relation between this distance and the rate of finding. In those experiments, the rate of contact with the eggs was determined for each parasite during the period in which it made contacts; only the time spent in. movement being counted. The number of contacts made per minute in each of the experiments is shown in Table III, and the average rate for each spacing is plotted against the distance between the eggs in Fig..

9 Host-finding by Insect Parasites Table III. The effect on the rate of finding hosts of variation in the distance between them, shown as number of contacts per minute Exp. - o-i Distance between eggs (in.) - O- - i I. Total Average II i ' - H i ' ' o ' - i- u. u a I o u I -" -" -" -" -" -" Distance between eggs Fig.. The effect on the rate of finding hosts of variation in the distance between them. The results show that the rate of finding eggs varied inversely as the distance between them. In other words, the parasite found hosts more quickly when they were near together than when they were far apart. It appears at first sight that such a conclusion was entirely to be expected, but further consideration shows that this is not so. In all the experiments with eggs spaced at -, o-i, - and - in., the Trichogramma was confined within a Petri dish on a surface bearing eggs. The area actually occupied by eggs varied in the

10 J. LAING four series of experiments, but in all cases the parasite was free to move over th whole of the paper underneath the dish and over the glass sides and top of the latter, a total area of - sq. in. The concentration of hosts in the area available for movement was, in fact, the same in the four series of experiments. Hence, if the parasite had been a particle moving at random, its average rate of contact with the eggs, measured over.a sufficiently long period, would have been constant, irrespective of the distance between the eggs. Since the rates obtained in the experiments vary profoundly with the distance between the eggs, it must signify that the Tnchogramma does not move at random about the Petri dish but restricts itself to the vicinity of the eggs. Two factors produce this restriction. One is, that by moving to an egg which it has seen, a Trichogramma brings itself into a more favourable position for finding the next egg. The other, and more important, faotor is the turning movement which follows an oviposition or examination, and which tends to make the parasite return to the neighbourhood of the eggs and so to avoid the barren spaces in the area available for movement. In order to appreciate this effect more fully it is necessary to obtain some idea of the size of the area to which the Trichogramma confines itself when moving among its hosts. For this purpose calculation was made of the size of the areas of movement in which a parasite moving at random would touch the eggs at rates equal to the rates of contact obtained in the experiments. A formula given by Stanley ( X, p. ) for the calculation of the probable number of contacts between an insect moving and eggs stationary in a limited volume of flour was modified for this purpose. The number of contacts in unit time between a moving particle and v stationary particles is G where G represents the volume of the flour (i.e. the volume of the space through which the particles are distributed, less the volume occupied by the particles); v the number of stationary particles; a x the mean diameter of the stationary particles;ct B the mean diameter of the moving particle; and /x the mean speed of the moving particle. Modified to apply to the movements of a particle over a surface bearing eggs, the number of contacts in unit time is -^ where G represents the total surface available for movement; v the number of Sitotroga eggs used in an experiment (i.e. ); fj. the mean speed of Trichogramma at about C. (- in. per sec, or (- x ) in. per min.); and (ffi + o-a) tne width of the area of perception of a Trichogramma for a Sitotroga egg (i.e. - in.). Using this formula, iv contacts per minute would result from random movement of a particle in an area G where G = flv ^ *~\ so that in the experiments with eggs - in. apart (- x ) x T.O x - G = sq. M in. = i-isq. in. -

11 Host-finding by Insect Parasites Similarly, the values of G obtained' by substituting the other five values of N would be: For o-i in. distance between the eggs, G= - sq. in. -..» G== -» - G= -,, - G=n-» - G= - It follows that the rates of -,, -, -, - and i- contacts per min. would occur with random movement in areas, respectively, of -, -, -, -, - and - sq. in. Now it is clear that these areas, which may be called the "areas of search", bear no relation to the size of the surface available for movement, since this is constant at - sq. in. for each of the egg-spacings - to - in. It can be seen from Table IV, however, that they show some approximation to the size of the areas over which the eggs are distributed, the area of search approximating more closely to the area of distribution of the eggs when these are close together than when they are farther apart. Table IV. The relation of the area of search to the area of distribution of the eggs Distance between eggs (in.) - O-O - - Area of* distribution (sq. in.) ' - ' rt Area of search (sq. in.) - - * - / Area of search Area of distribution i S The small area over which the eggs are distributed is in the form of a rectangle ( xd) in. long and (o' + rf) in. wide, where d = distance between the eggs. Hence the significance of the relation between the rate of finding and the distance between the hosts is that a Trichogramma does not seek at random over the whole of a field of movement but virtually restricts its search to an area approximating in size to the small area over which the hosts are distributed. In brief, the process of finding hosts by Trichogramma is more efficient than it would be if its movements were those of a particle moving at the same speed at random. THE EFFECT UPON. FINDING OF VARIATION IN THE SIZE OF THE HOST The relation of the frequency of neighbouring contact to the size of the egg was investigated with eggs of four sizes: (i) selected small eggs of Sitotroga (average length - mm., average breadth - mm.); () selected large eggs of Sitotroga (o-i x - mm.); () eggs of Hydriomena bilineata (- x - mm.); and () eggs of Euxoa segetum for eight experiments and of Barathra brassicae for two experiments. The eggs of Euxoa and Barathra were very much alike, subspherical, and of average diameter o-i mm., average height - mm.

12 J. LAING Ten experiments were carried out with eggs of each size, in daylight and at - C. The procedure was the same as that described on p., except that in all the experiments the distance between the eggs was - in. A slight modification of the method was necessary in the experiments with Hydriomena eggs, for their shell was so hard that Trichogramma females took between and min. to oviposit, they frequently oviposited repeatedly in a single egg without moving away between one oviposition and the next and, moreover, after parasitizing eggs for some time, many females stopped moving altogether and remained continuously on one egg for periods of or min. It was, therefore, not always possible to watch each Trichogramma until she had made contacts, and so, to compensate for the shortness of most experiments, observations were made on twelve females, and, in two cases in which the female had not stopped moving, observations continued up to the th contact. The results of the experiments are shown in Tables V and VI. Table V. The effect on the frequency of neighbouring contact of variation in the size of the host Small eggs of Sitotroga Large eggs of Sitotroga Exp. Ovip. No. of Exam. Touch NC after Ovip. and exam. Touch Total NC Ovip. No. of Exam. Touch NC after Ovip. and exam. Touch Total NC JO Total S ' S " IS Eggs of Hydriomena Eggs of Euxoa and Baratltra S II Total no S S i H ' ' Small eggs of Sitotroga NC/C% -; NCJC.% -; NCJC t % -. Large eggs of Sitotroga NC/C% -; NCJC /o -; NC,/C,%. Eggs of Hydriomena NC/C% -; NCJC O.%%TT, NC t /C,% -- Eggs of Euxoa and Barathra NCjC% -; NCJC C -; NCJC t %.

13 Host-finding by Insect Parasites Table VI. Effect on the rate of finding hosts of variation in their size Exp. Small eggs of Sitotroga c.p.m. Climbs Large eggs of Sitotroga c.p.m. Climbs Eggs of Hydriomena c.p.m. Climbs Eggs of Euxoa and Barathra c.p.m. Climbs Total Average - 'O ' ' -O 'i S'S -S - - i'. ' S "S ' S - Two notes of explanation are required in connexion with these results. () The parasite frequently oviposited two or more times in a single egg of Hydriomena, Euxoa and Barathra without leaving the hosts. Since the number of ovipositions in these experiments is recorded chiefly for its value in indicating the proportion of times the parasite made a turning movement when leaving a host, the figure given for the ovipositions is the number of times the Trichogramma left an egg in which it had oviposited. () The number of times the parasite climbed on the Petri dish during an experiment is an index of the extent to which it wandered from the immediate vicinity of the hosts. Those insects which remained close to the eggs throughout the whole time of an experiment never, or rarely, climbed upon the glass sides of the dish, whereas those which left the egg area between ovipositions frequently did so. The frequencies of neighbouring contact, total, after oviposition and examination, and after touch, are plotted against egg size in Fig.. The sizes of the eggs are taken as - + h for the Sitotroga and Hydriomena eggs and as (d + h) for the Euxoa and Barathra eggs, where / = length, b = breadth, h = height and d= diameter. The comparable egg sizes according to this calculation are, for small eggs of Sitotroga * mm., for large eggs of Sitotroga - mm., for Hydriomena eggs - mm., and for Euxoa and Barathra eggs - mm. From inspection of the tables and the graph it is clear that the frequency with which Trichogramma makes a neighbouring contact varies directly with the size of the egg. How variation in size affects finding can be understood by consideration of the above results in detail. It is obvious that even if Trichogramma could not see the eggs, increase in their size would result in increase of neighbouring contact frequency. The parasite does see the eggs however, and, moreover, should see them from a distance which is proportional to their size. Thus, in order to determine how far

14 J- LAING the neighbouring contact frequencies obtained in the experiments are an effect of chance, it is necessary first to calculate the radius of perception for one particular egg size (i), and then, from this radius, to work out proportional radii of perception for the other egg sizes (). With these values it will then be possible to calculate the frequencies of neighbouring contact which would result from random movement () and to compare the calculated with the observed frequencies (). () To calculate the radius- of perception of an egg of Euxoa or Barathra, again, as on p., one must assume that for one value of the varying factor the frequency of neighbouring contact following touch is a result of entirely random movement on the part of the Trichogramma. In the present case this assumption is to be made for I r o a) I g o.s? e - O fr fa ot- ( lt - l' Size in mm. Small Large Hydriomena Euxoa Sitotroga Sitotroga and Baratlira Fig.. The effect on the frequency of neighbouring contact of variation in the size of the eggs. O O NCJC M %, NC/C %, X x NCjC t %, calculated chance of JVC. the frequency of neighbouring contact following touch obtained "with Euxoa and Barathra eggs. The justification for the assumption is, that when these eggs are spaced - in. apart, the parasite is not able to see one egg from its neighbour, nor yet is it able to turn much on its track between leaving one egg and perceiving a neighbour. Moreover, so far as can be judged from observation, Trichogramma never rejects Euxoa or Barathra eggs from a distance as it sometimes rejects Sitotroga eggs (see p. ). It seems reasonable to suppose therefore, that the frequency of neighbouring contact following touch obtained in the experiments with these very large eggs is a value dependent entirely on chance. Calculating from the value NC t /C t % = - (Table V), and referring to Fig.,.,,.,,..,. LBAC the chance of a neighbouring contact with one egg is -

15 Host-finding by Insect Parasites Therefore L BAC= ' S * =- ; and IOOX X C (radius of perception) = sin L - x ^, which, since AB is - in., =- in. () The distance from which eggs of Hydriomena and Sitotroga can be seen can now be calculated on the assumption that the distance from which an egg is seen is proportional to its size. Since Euxoa and Barathra eggs, with a size of i- mm., are perceived from the limits of a circle of radius - in.; Hydriomena eggs, with a size of - mm., are perceived from - / o \ ^ m- x (o-ei x -) in. = o-i-ji; in. (-x-) v JD Similarly, large Sitotroga eggs are seen from o- in. away and small Sitotroga eggs from - in. away. () It is now possible to calculate the frequencies of neighbouring contact that would occur with random movement. It has already been shown that a radius of perception of - in. corresponds to a neighbouring contact frequency of -% when the eggs are - in. apart. Using the method shown in detail on p., the chances of contact for Hydriomena eggs, seen from - m -> large eggs of Sitotroga, seen from o-iiin., and small eggs of Sitotroga, seen from - in., are calculated to be, respectively, -, - and -%. These calculated values are plotted in Fig.. () Comparing the calculated and observed frequencies, it is seen that those calculated for Hydriomena and large Sitotroga eggs are very close to those observed for neighbouring contact following touch. One may therefore conclude not only that the primary assumption was justifiedthat in the experiments with Euxoa and Barathra eggs the observed frequency of neighbouring contact following touch is a value dependent entirely on chancebut also that the corresponding frequencies obtained in the experiments with eggs of Hydriomena and large eggs of Sitotroga likewise depend entirely on random movements. Except for the experiments with small eggs of Sitotroga, the total frequency of neighbouring contact observed is greater than that expected from chance. It is clear that again the turning movement which follows oviposition and examination is responsible for this higher frequency, and that again the effect of the movement is differential. The effect is greatest for Hydriomena eggs (~ : ^-' = - : ) \ C M C t INC NC \ and least for Euxoa eggs (--, *: -~-*= -:). It is probable that with eggs as large as Euxoa eggs, spaced as close as - in., the turning movement is wasted, as it is with Sitotroga eggs when they are o-io in. apart. With small eggs of Sitotroga, however, the observed value of NC t /C t is only -% while the calculated value is -%. The parasite, that is, finds the small eggs less often than one would expect. This discrepancy can be accounted for by consideration of the attitude of the parasite towards the small eggs. It was noticed

16 J. LAING on several occasions that a Trichogramma approached very close to a small egg and went away again without touching it. In these cases, it appeared that the parasite saw the egg but rejected it from a distance. This observed rejection was not the only sign that the small eggs were less acceptable than the large eggs for, in the ten experiments with small eggs of Sitotroga, only eggs were parasitized whereas, in the three series of experiments with the larger eggs,, and eggs were parasitized. The rejection of these small eggs of Sitotroga, either before or after contact, is a normal part of the behaviour of the parasite, for Salt (, p. ) has shown that "the principal criterion used by ovipositing females of Trichogramma in the selection of their hosts is that of size". When, in the experiments with small Sitotroga eggs, the parasites found the neighbouring egg less frequently, they tended to move away from the vicinity of the eggs. Table VI shows that the number of times the parasite left the egg area and climbed on the Petri dish varied inversely with the size of the egg, and was disproportionately large in the experiments with the small eggs of Sitotroga. When the Trichogramma had climbed on the glass, it appeared as though it had "lost interest in" or "desire to find" the eggs, for frequently it stayed on the glass at the side nearer the window for considerable period? of time. On three occasions experiments with small eggs were started but had to be abandoned, since the parasite, after finding and rejecting several eggs, wandered away to the light and did not return to the egg area. Thus, when the hosts are very small, the direct effect on finding that must inevitably be exerted by their small size, is exaggerated by the effect of their rejection as unacceptable hosts. This effect is a disproportionate reduction in the frequency of neighbouring contact, in the rate of contact, and in the amount of time spent moving in the vicinity of the eggs. Occasionally this reduction amounts to a cessation of searching. The main conclusions to be drawn from this section concerning the effect upon finding of variation in the size of the host, are as follows: () The variation in the frequency of neighbouring contact as the size of the host is altered can be accounted for chiefly as a mechanical effect of size on the chance of a contact. () The direct inechanical effect of size is modified in two ways. First, the turning movement following oviposition and examination increases the frequency of neighbouring contact with eggs of all four sizes, but increases it disproportionately. Secondly, rejection of very small eggs decreases the frequency of neighbouring contact with them. The total effect is that Trichogrammafindslarge hosts more efficiently in proportion to their size than it finds small hosts. HOST SELECTION AND HOST-FINDING Since if other things are equal there is more chance of finding large than small eggs, it is likely that when eggs of two sizes are present in the same area of movement there is some degree of host selection by selective finding. Moreover, judging from the observations on the effect of size, it is probable that the extent of the

17 Host-finding by Insect Parasites selection by finding will depend upon the distribution of the two kinds of eggs relative to one another. A small degree of selective finding was shown by Salt (, p. ) with large eggs of Ephestia and small eggs of Sitotroga, arranged alternately o-io in. apart in a in. square on graph paper. In the course of four experiments each with eggs exposed to one female Trichogramma, he observed contacts with the Ephestia and with the Sitotroga eggs. Dr Salt kindly allows me to make use of his notes of three other sets of experiments, of the same type as these but with different combinations of eggs. The actual numbers of contacts with the eggs, the dimensions of the latter, the size taken as comparable standard of measurement, and the ratios of contacts and of the sizes of the eggs in each of the four series of experiments are given in Table VII. Table VII. Selective finding when eggs of two sizes are arranged alternately in one group Eggs used Dimensions (mm.) Standard size (mm.) Contacts Size ratio Contact ratio Agrotis v. small Sitotroga Agrotit v. small Ephestia Large Ephestia v. small Sitotroga Agrotis v. large Ephestia -x- - X X x - X -x- -x-x- -x- xo- - x - -x-x- o- O- - - to to to to IOO J' For each series the ratio of the contacts made by the Trichogramma with the two kinds of eggs is seen to be almost equal to the ratio of the egg sizes. Where there is a discrepancy it is to the effect that more contacts are made with the smaller eggs than are to be expected from their relative size. The probable explanation of the discrepancy is that, since the parasite oviposits more often in the larger eggs, the chance of contacts with their neighbours is increased; that is, the chance of hitting thesmall eggs, which are arranged alternately with the large eggs, is increased. Whether or not this explanation is correct, the similarity of the contact and size ratios is in all cases sufficiently close to make it clear that the relative numbers of contacts with the two kinds of eggs depend upon their relative sizes. The selective finding which has taken place in these experiments is, thus, a direct and automatic effect of the difference in size of the two kinds of eggs. It has been shown that very small hosts, which are frequently rejected by Trichogramma, are found less easily in proportion to their size than are large and acceptable hosts. Although in the preceding experiments on host-selection the small hosts were frequently rejected, their arrangement alternate with the large hosts allowed only the direct effect of their small size to appear. The question now arises whether the indirect effect of small sizethat is, the effect of the rejection of very small eggscan show itself in the degree of selective rinding of eggs of two jeb-xviii

18 J. LAING sizeslarge and acceptable, and small and much less acceptablewhen these are arranged, not alternately in one group, but in separate groups. Experiments to answer this question were carried out as follows. Selected large and small eggs of Sitotroga were pasted on graph paper in six small groups, with a distance of -o in. between the centres of neighbouring groups. In each group were nine eggs, either all large or all small, arranged o-io in. apart in a square. The three groups of large eggs alternated with the three groups of small eggs, and all were so placed that they fitted underneath the central part of a Petri dish. An experiment consisted in noting the movements of a female parasite among the eggs until she had found one or two groups of both kinds of eggs. The results of the experiments are given in Table VIII, of which one part shows the contacts made with eggs in each group visited for the first time and the other shows the contacts made in groups visited for the second, third or fourth time. Table VIII. Selective finding when eggs of two sizes are arranged in separate groups Numbers of contacts and ovipositions and examinations of large and small eggs First visit to group Repeat visits to group Exp. Large eggs Small eggs Large eggs Small eggs Ovip. Exam. Ovip. Exam. Visit Ovip. Exam. Visit Contact Contact Contact Contact Ovip. Exam. i S IO Total IS O f. u * in IS IS nd nd rd th nd rd nd nd rd th nd nd nd rd nd rd nd nd nd nd nd rd th nd IS IS. IS IS i

19 Host-finding by Insect Parasites It is necessary to discuss first the results shown in the left-hand column, for from them the main question of the experiment can be answered. The average number of contacts made with the large eggs in a group visited for the first time was -, the average number made with small eggs was IO-I. Since the ratio of contacts is / (- : ) whereas the ratio of egg sizes is o-/o' mm. (*: ), the selective finding of large eggs cannot have been merely a mechanical effect of their greater size but must have resulted from a difference in the behaviour of the parasite with respect to the two types of host. In most cases, when the Trichogramma encountered a group of large eggs, each of the first few contacts was followed by an oviposition, and contacts which remained merely touches were infrequent until several of the eggs had been parasitized. In a few cases the small eggs were readily acceptede.g. in Exps., () and ()and the parasites remained on the groups; but generally a large proportion of them were rejected and the first few contacts resulted in fewer ovipositions and more touches. Hence, since the chance of a contact with a neighbouring egg is smaller after a touch than it is after an oviposition, there was more chance of the parasite leaving a group of small eggs after making only one or two contacts than there was of it leaving a group of large eggs: In these experiments, therefore, selective finding out of proportion to the size of the eggs occurred as a result of the rejection of very small eggs. So far only the first visits to the groups of eggs have been considered. Since the parasite was confined within the limits of a Petri dish, it frequently happened during the experiments that groups of eggs were encountered more than once. In order to determine whether the parasite continues to find more of the large than of the small eggs when the area available for search is small, it is necessary now to consider the repeat visits to the groups of eggs. In the course of ten experiments, the parasite times encountered groups of eggs which had previously been met and in which one or more ovipositions had been made. The total number of contacts made in twelve repeat visits to groups of large eggs was sixty-four, and to groups of small eggs ninety-one. In most cases, when a group of large eggs was encountered for the first time, almost all the eggs were parasitized before the group was left; hence, when the parasite encountered a group of eggs for the second or third time, being able to discriminate between healthy and parasitized eggs (Salt, b), it merely touched and rejected them. As a result, very few contacts were made. In the case of the small eggs, on the other hand, only a small proportion were parasitized at the first visit and the Trichogramma was able at later visits to examine and occasionally to oviposit in the previously untouched or rejected eggs. Thus, when a parasite has been searching for some time in a small area, in its second and subsequent encounters with the groups of eggs it finds a greater number of the small and unparasitized eggs than of the large eggs which were parasitized at the beginning of the search. In brief, the two series of experiments with large and small eggs together, show that selective finding of large eggs occurs to some extent when the two kinds of

20 J. LAING eggs are arranged alternately, and to a greater extent when they are arranged in separate groups. In the latter case the selective finding of the large eggs continues only so long as the parasite encounters groups of large eggs which have not been found before. DISCUSSION It is obvious that the success of a parasite in finding its hosts depends upon many things, some of them connected with the parasiteas, for instance, its powers of movement and perceptionand some of them connected with the hostssuch as their size and distribution. In a previous paper the nature of the movement and perception of Trichogramma was dealt with; in this paper, an attempt has been made to determine the influence on finding of variation in the size and distribution of the host. It was clear at the outset, that by altering the size of the hosts or the distance between them, the parasite's chance of finding the hosts would be altered, and it is evident from the results of the investigation that the observed variation in the frequency of finding can in fact be accounted for chiefly by this variation in chance. The experiments show, however, that the direct effect of each factor upon the chance of contact is modified by the behaviour of the seeking parasite, so that hostfinding is only partially and not entirely a matter of chance. For instance, by reason of the turning movement which follows parasitization or examination of a host, eggs are found more readily than they would be by random movements. Since, however, the finding efficiency of Trichogramma, as measured by the ratio area of search: area of egg distribution, is least when the eggs are - in. apart, the effect of the turning movement is less marked when the eggs are sparsely distributed than when they are spaced more closely. This being so, it might be claimed that the peculiar way in which Trichogramma moves from its hosts makes it more likely to occur in nature as a parasite of eggs laid in groups than of eggs which are scattered singly. Further, since turning does not occur when the Trichogramma is leaving a host which it has merely touched, when eggs are encountered which are not acceptable hostssuch as eggs which are too small or which are already parasitizedthe parasite does not restrict its movements to their neighbourhood, and so saves itself from repeated useless encounters. In connexion with' its finding efficiency there are two points in the behaviour of Trichogramma which are particularly importantthat a female parasite is able to reject hosts that have already been parasitized by another Trichogramma (Salt, J) and that it searches more efficiently around groups of eggs that are acceptable than around those that are not acceptable. Together these facts make it probable not only that individual searching is not random, but also that the searching of populations of this parasite is not entirely random. This matter is to be investigated in future experiments. In this paper only two factors have been studied, the size and the distribution of the hosts. It is to be borne in mind, however, that host-finding must also be

21 Host-finding by Insect Parasites influenced by environmental conditions such as light, temperature, or humidity, factors which are external to the parasite and host but which affect the behaviour of the parasite. That variation in light-intensity, for instance, affects the frequency of neighbouring contact by its influence on the distance from which the parasite sees its hosts is indicated by the following figures. In experiments with Sitotroga eggs, chosen at random and spaced - in. apart, the frequencies of neighbouring contact following touch were: (i) in daylight in June and July, '%; (ii) in October and November, -%; (iii). in late-november and December, -%; (iv) in a room lit only by a W. lamp, - %; (v) in the same room lit by a W. lamp, - %. A difference in the ability of Trichogramma to see Sitotroga eggs in summer daylight and in artificial light was quite apparent by mere observation. Moreover, in very bright light Trichogramma moves about restlessly and appears to lose interest in its hosts, and often in darkness it does not seek at all. Temperature also has these two effects, for within certain limits, variation in temperature influences the rate of finding by reason of its effect on the rate of movement of the parasite; and again, beyond these limits the parasite becomes so restless or so inert that it will no longer seek hosts. External factors such as light and temperature, therefore, while influencing finding in proportion to their intensity for part of their range of variation, when at very high or very low intensities affect the condition of the parasite in such a way that host-finding no longer occurs. It is evident from this consideration not only that success in finding hosts by Trichogramma depends largely upon the behaviour of the parasite, but also that the behaviour varies considerably, with varying external conditions. Once again the behaviour of the parasite intervenes and prohibits any purely mechanical treatment of its interaction with its host. SUMMARY. Experiments are described in which Trichogramma females were provided with Sitotroga eggs arranged at various distances from - to - in. The frequency with which the parasite found the neighbouring egg varied inversely with the distance between the eggs. The relation between frequency of contact and distance between the hosts depends chiefly, but not entirely, on chance.. The rate of finding the eggs varied inversely with the distance between them. By consideration of the conditions of the experiments, this relation between the rate of finding and the distance between the eggs is seen to indicate that the Trichogramma does not seek at random over the whole area available for movement but restricts its search to the neighbourhood of hosts.. From experiments with eggs of four sizes it is shown that frequency of neighbouring contact is correlated positively with the size of the eggs. The influence of size is largely mechanical; but not entirely, for small eggs of Sitotroga are found less efficiently in proportion to their size than are larger eggs.. When provided simultaneously with hosts of two different sizes, a Trichogrammafindsmore of the larger than of the smaller hosts. If the hosts are arranged

22 J. LAING alternately, the amount of selection that occurs by finding depends entirely upbn the relative sizes of the eggs. If they are arranged in separate groups, however, relatively more of the larger eggs are found. ACKNOWLEDGEMENT I am deeply indebted to Dr George Salt for the use of his culture of Trichogramma, and for his most valuable- help and criticism throughout the course of this work. REFERENCES LAING, J. (). J. Amm. Ecol.,. SALT, G. ( a). Proc. roy. Soc. B,,. (). Proc. roy. Soc. B,,. (). Proc. roy. Soc. B,,. (). Proc. roy. Soc. B,,. STANLEY, J. (). Canad. J. Res., a.