On Intraspecific and Interspecific Competition in Larvae of Graminivorous Insects

The behaviour of the beetle, Rhizopertha dominica, is adapted to the needs of its larvae in that the sensations which induce the females to oviposit would usually lead to eggs being laid only in environments where sites suitable for larval development are to be found (Crombie, 1942). But the rate of oviposition, even at the extremes of adverse conditions possible in actual populations, is so great that a considerable mortality must occur in the immature stages before the offspring are reduced to numbers which the environment could support (Crombie, 1942). The females of the moth, Sitotroga cerealella, appear likewise to have little control over their rate of oviposition (Crombie, 1942). Their ability to select, under natural conditions in the field, an oviposition site suitable for larval development is imperfectly known (King, 1918; Simmons & Ellington, 1927, 1933), but in a confined space it seems adequate (Simmons & Ellington, 1924; Crombie, 1942). The females of neither species avoids ovipositing on already infested grains. In the course of the experiments on oviposition already published (Crombie, 1942) it was frequently noticed that the eggs of either species were deposited next to or even on top of previously oviposited eggs of the same or of the other species (cf. Ullyett, 1936; Salt, 1937; Lloyd, 1939, 1940). Wheat grains already containing larvae of either species also seem to be as acceptable to the females of both species as uninfested grains. Even if the gravid females did avoid ovipositing on previously infested grains her efforts would be vitiated both by the size of the egg batches often placed on one grain, and by the habit of the active first instar larvae of both species of wandering away from the place where they have hatched. The rudimentary nature of’ parental care’ in these two species of insect means that the larvae must be adapted, particularly during the early instars, to an independent existence. This will expose them, in choosing and occupying situations in which to complete their development, to the competition of other larvae similarly engaged. In this paper the term superinfestation will designate the cases in which one wheat grain is inhabited simultaneously by two or more young of the same species of insect. The term multiple infestation will designate the cases in which one wheat grain is inhabited by the young of two or more different species of phytophagous insect (cf. Smith, 1929). It is important to know firstly, whether the larvae themselves avoid multiple or superinfestation, and if so, by which senses they perceive the larva already in occupation of the grain; and secondly, if multiple and supennfestation are not avoided, how overcrowding affects their development and what is the outcome of competition between the two species in mixed populations (cf. MacLagan, 1932; Timofeeff-Res-sovsky, 1933, 1935; L’Héritier, 1937; Bodenheimer, 1938). Overcrowding may evidently affect living organisms through the limitation of food, oxygen or space (leading to reflex stimulation caused by the perception of other individuals, e.g. by actual contact), or the accumulation of excretory products (‘conditioning’ of the medium) (Allee, 1931, 1934; Hammond, 1938, 1939; Clements & Shelford, 1939; Park, 1939, 1941 ; MacLagan, 1941). When insect larvae are exposed to such conditions this may lead to the death of individuals before reaching maturity (MacLagan & Dunn, 1935; Salt, 1936); the reduction in size of the pupae and resulting adults (Weidling, 1928; Holdaway, 1930; Salt, 1932a), or of the vigour and longevity of the latter ; the lowering of the fecundity of the adult females (Hofmann, 1933 ; Dunstan, 1935); retardation (Landowski, 1938) or stimulation (Michal, 1931 ), of the rate of development ; changes in the sex ratio of the emerging adult population (Herms, 1928; Brandt, 1937; Flanders, 1939; Graham, 1939); or failure to make full use of the food reserves of the medium because of competition for space. The latter may lead to survival being lower at high than at low densities.

The experimental conditions in these experiments were the same as those described in a previous paper, to which reference should be made (Crombie, 1942). All the experiments were performed in a dark incubator at 30°C. and 70% R.H.

The larvae of both species usually bore into a suitable object (e.g. a wheat grain) in the first instar, and generally remain there, feeding upon the interior of the grain, until they have completed their development (Barnes & Grove, 1916; Fletcher, 1920; Back, 1922). There are usually four larval instars in Rhizopertha followed by a prepupal stage which does not involve moulting. This moults to form the pupa, which is followed by the adult. The larval instars are with practice easily distinguished by eye by means of the characteristics described by Potter (1935), but were checked by measuring the head capsule. Under our experimental conditions the average duration of the instars was as follows: egg, 6 days; first larval instar, 6 days ; second, 6 days ; third, 5 days ; fourth, days; prepupa, 1 day; pupa, 7 days; total, 35 days. In Sitotroga there are also usually four larval instars, which are with practice readily distinguishable by eye. As before, they were when necessary checked by measuring the head capsule. The average duration of the instars under our experimental conditions was as follows: egg, 3 days; first larval instar, 6 days; second, 6 days ; third, 5 days ; fourth, 7 days ; pupa, days; total, 32 days (Crombie, 1943). The period from hatching to emergence was thus 29 days for each species. This period was sometimes abnormally extended (vide infra). In both species the pupa is found in the cavity made by the feeding larva. Escape from the grain by the adults of Sitotroga is accomplished by the act of emerging from the pupa. The adults of Rhizopertha usually remain inside the grains for a few days while their cuticle hardens before eating their way out. The larvae of Rhizopertha can enter intact wheat grains only in the first instar. This can be accomplished by the larvae of Sitotroga in all instars, the resulting mortality decreasing with age (Back, 1922). The mortality among first instar larvae of both species as a result of such an attempt may be as high as 50 %.

The results of all statistical tests in this paper are given in terms of the value of p, which represents the probability of obtaining as great a deviation by random sampling as that observed in the experiment. A probability (p) of 5 % is usually taken as the arbitrary division between significance and nonsignificance, so that when p<0·05 the difference between the values compared may be regarded as significant (Simpson & Roe, 1939).

Single larvae were confined in cells made by gumming a cover-slip over a glass ring. A series of such cells was stood on a ground-glass plate. Each cell contained besides the larva two objects, each composed of different substances, between which the larva had to make a choice of which it was to enter. After 6 hr. in the incubator the cells were examined and the positions of the larvae noted. The latter always bored into one or other of the objects. Twenty Rhizopertha larvae of each instar were offered the choice of each of the following pairs of substances : a piece of wheat grain and a piece of cork cut to the same shape and size ; a false grain made of plaster of Paris mixed with wheat flour and a false grain made of plain plaster of Paris (Crombie, 1941). In both experiments all twenty larvae of each instar chose the wheat. The distinction made between the false grains of plaster containing wheat flour and those composed of plaster alone suggests the existence of a chemoreceptive sense, since the only difference between them is provided by the food. The first experiment was repeated with twenty Sitotroga larvae of each of the four instars, and in each case the wheat was preferred to the cork twenty out of twenty times. Twenty first instar larvae of each species, respectively, were now given the choice of a piece of bean seed and a piece of cork of the same shape and size, and of a piece of wheat grain and a piece of bean seed of the same shape and size. All the Rhizopertha and eighteen of the Sitotroga larvae chose the bean to the cork; seventeen of the Rhizopertha and all of the Sitotroga chose the wheat to the bean. The larvae of all instars of both species can thus recognize objects containing food, and the first instar larvae of both species are able to distinguish between different kinds of food.

As already stated, the proportion of first instar larvae of either species which manage to enter intact wheat grains may be as low as 50 %. With such a high mortality the experiments to be described in this paper could not have been performed. It was necessary therefore to damage the test in some standardized and repeatable way, so that the larvae could enter more easily. This was done by gently tapping the grains until the test was just broken by a crack. The larvae almost invariably entered the grains in the damaged regions and mortality during entry was reduced almost to zero. The grains were treated in this way in all the experiments described in this paper. This may appear to introduce an unnatural condition, but we are chiefly interested in the competition inside the wheat grains, and not in mortality which is the result of the accident of their possessing a hard test. This study must be regarded, however, as a study in competition for ‘cracked grains’, and not for intact wheat.

150 wheat grains were placed into each of three dishes so that they were one layer deep. Into one of these dishes (a) were introduced 100 first instar Rhizopertha larvae, into a second dish (b) 100 first instar Sitotroga larvae, and into the third (c) 50 first instar larvae of each species. After 24 hr. in the incubator the grains were dissected and the positions of the larvae noted. The results are given in Table 1. The number of grains containing o, 1, 2, 3, etc., larvae, assuming that the latter distribute themselves at random among the grains, were calculated from the Stoy formula where x is the number of larvae and N the number of grains (Appendix to Salt, 1932b). Each of the observed values (a), (b) and ((c) : total) were compared with these calculated values by means of χ², and none of them was significantly different from the latter (p > 0·05). In (c) the members of each species entered the grains in practically equal numbers. It may be concluded therefore that the larvae of neither species avoid either superinfestation or multiple infestation (cf. King, 1918; Simmons & Ellington, 1925 ; Bodenheimer, 1930). Each species, however, possesses a kind of behaviour by which overcrowding may be prevented, viz. migration. The habit of migration of Sitotroga larvae from the grains they were inhabiting has been remarked upon by Back (1922), but no reference has been found to it m Rhizapertha.

The non-avoidance of superinfestation by Rhizopertha larvae was confirmed by the following experiment: each of a number of first instar larvae was given the choice between a fresh grain and a grain (marked with a spot of ink) which contained three living larvae. The two grains were placed side by side on a ground-glass plate, and covered by one of the cells described above. The larvae cannot escape from such cells, a series of which were arranged on each glass plate. After-6 hr. the grains were dissected by slicing with a sharp scalpel under a binocular microscope, and the positions of the larvae noted. In approximately half the cells (23 out of 50) the fresh grains contained one larva while in the other half (27 out of 50) the infested grains contained four larvae, showing that half the larvae had entered each set of grains. There was therefore no avoidance of the infested grains. The point to be decided now was whether or not it is the presence of other larvae in the same grain which causes a larva to migrate. Fifty first instar larvae were placed separately under cells with one wheat grain each, and the whole put into the incubator for 6 hr. The cells were then examined and all the larvae were found to have entered the grains. A fresh grain (marked) was now introduced into each cell and the whole returned to the incubator once more. Every 2 days until adults emerged the fresh grains were removed, dissected and replaced by other fresh grains. No larvae were found in the fresh grains and eventually forty-two adults emerged from the infested grains. Those from which adults had not emerged were then dissected: dead first instar larvae were found in four of them, dead larvae of later instars in two more, while the other two contained dead, somewhat deformed adults. It may be concluded then that migration occurs only when two or more larvae enter the same grain. Mortality may occur, however, in larvae occupying grains singly.

If larvae migrate thus owing to the influence of other larvae occupying the same grain, it may be asked by what sense a larva perceives that there is another larva present. There are three possibilities : chemoreception, ‘hearing’ or stimulation from mechanical vibrations (cf. Minnich, 1925) and stimulation resulting from actual contact. Now if larvae migrate after detecting the presence of other larvae by chemoreception, freshly killed larvae ought perhaps to cause migration in the same way as living larvae. A number of freshly killed larvae were introduced into a hole made in a wheat grain, a living first instar larva placed beside each such grain, and both covered with a cell. After 6 hr., during which the larvae entered the grains, a fresh grain (marked) was introduced into each cell. After 48 hr. the grains (50 in all) were dissected and all the larvae were found still in the grains containing the dead larvae. Chemoreceptive or other stimuli from the latter do not therefore cause migration.

The effect of mechanical vibrations was investigated as follows. Two grains, each containing three first instar larvae, were firmly cemented together. Fifty such pairs were set up. Each of these, together with five fresh grains to each, was then placed on a glass plate and covered with a cell. After 2 days the grains were dissected and the positions of the larvae noted. Now as described below, the number of larvae which migrate from or die in a grain increases with the initial number present (Table 2). Mechanical vibrations are not likely to cause death, so we may omit the dead larvae for the moment. If such vibrations were the cause of migration one might expect a similar increase in migration with an increase in this factor. Now one would expect that the amount of vibration or noise produced by six larvae in two grains cemented together, would be greater than that produced by three larvae in a single grain. However, the average number of larvae migrating per grain from the cemented grains was 0·32, which is not significantly different from the number, 0·31 per grain, of larvae migrating from single grains each containing initially three larvae. There was also no significant difference between the total number of larvae killed and migrating per grain in the two experiments, this value being 0·4 and 0·38 for cemented pairs and single grains respectively (Table 2). Mechanical vibrations do not therefore seem to be the cause of migration.

The probable importance of actual contacts between larvae as a cause of migration becomes immediately apparent. The reaction of one larva to the presence of another was observed as follows. When two larvae in the first, second, third or fourth instars were put together into a small hole drilled in a wheat grain and watched under a binocular microscope, they were often seen to attack each other with their mandibles, and eventually either one or both left the hole. When a larva entered such a hole it always went to the bottom and turned round so as to face outwards. Other larvae trying to enter the hole were fiercely attacked. Sometimes such combats resulted in the body wall of one of the antagonists becoming punctured and its bleeding to death. In their tunnels in wheat grains larvae of all instars were always found curled up with the head facing towards the way they had entered. Furthermore, in all grains dissected during the experiments to be described, whenever two larvae were found in the same tunnel at least one of them was always dead. It thus seems probable that whenever two larvae meet within a grain they will attack each other, with the result that either or both will migrate or be killed.

Different numbers of larva were introduced into wheat grains as follows : the requisite number of larvae was placed with a wheat grain on a ground-glass plate and covered with a cell. A number of such cells were set up. After 6 hr. in the incubator the grains were examined. All the larvae had usually entered them by this time, grains in which they had not done so being rejected. Five fresh grains (each marked with a spot of ink) were now introduced into each cell and the whole returned to the incubator for 48 hr. At the end of this period the grains were all dissected by slicing with a sharp scalpel (the original infested grains being examined first in order to minimize errors due to the extension of the 48 hr. period by the time taken in dissection) and the positions of the larvae noted. This provided data about the number of larvae killed in and migrating from each original grain during a constant period of time. This procedure was followed with larvae in the first, second, third and fourth instars. The first instar larvae had recently hatched, the others having recentiy moulted. The results are given in Table 2. With initially only one larva per grain no migration or death occurred during this 2-day period. During this period of their lives both these effects were therefore apparently due to the presence of more than one larva in each grain, and the increase in migration and death with increased initial numbers was due to crowding.

Two further points of interest in Table 2 may be noted. First, as larval density increases the proportion of larvae killed as a result of encounters increases, while the proportion migrating decreases (column 8). Secondly, the number of larvae killed and migrating per grain increases with later instars (column 5). This may be because the larger size of later instars increases the probability of encounter. Finally, it may be mentioned that the number of larvae killed and migrating from grains which had been cut either in half or in quarters was significantly greater than from the whole grains. These observations are of interest merely in that they confirm the foregoing results, so that no details will be given.

The existence of larval fighting in this species was observed nearly 200 years ago by Duhamel & Tillet (1762). These authors remark that they often found three or four dead larvae in grains of which only one living larva had taken possession. On one occasion two larvae were seen apparently fighting. After several minutes one was dead and the other was seen working its way into the gram. They were never able to find more than one living larva per grain. Similar observations have been made by Flanders (1933). Fighting, sometimes resulting in death, has been observed here, also, among larvae confined together in a small hole bored in a wheat grain. In all the grains dissected during the experiments described below two living larvae were never found in the same tunnel, although a Living and a dead larva, or two dead ones, were sometimes found together. The experiments described above with Rhizopertha were now repeated with this species, but only with first instar larvae. As the results were similar to those obtained with the former species, and negative, no details will be given. The larvae did not avoid entering grains already containing other larvae, and chemoreceptive and ‘auditory’ stimuli were, as before, apparently not causes of migration. An experiment to determine whether migration or death occurred in grains containing only a single larva was carried out as with Rhizopertha. No migration occurred, and from the fifty grains forty-three adults eventually emerged (cf. Barnes & Grove, 1916; Fletcher, 1920). On the dissection of those from which adults had not emerged four dead first instar larvae, one dead third or fourth instar larva, one dead pupa, and one dead, deformed adult were found.

The average reduction in numbers of larvae per grain by migration or death during a fixed period of time under different degrees of crowding was now determined as before. The results are shown in Table 3. Seven out of a hundred larvae died, during the 2-day experimental period, in the grains containing initially only one larva. This means that 7 % of the larvae died, without encounters with other larvae, in the first 2 days after hatching. There seems to be no reason why such mortality should not occur at any density, so that we must subtract from the total number of larvae killed and migrating at any density (column 4), 7 % of the initial number of larvae introduced into the grains. The remainder is the number of larvae killed and migrating as a result of random encounters (column 5). The value of k was calculated as before from the observed value of p_{ when i = 2, and in this case k = 0·09. The values of p_i/i (expressed as percentages) calculated from this value of k at different values of i are given in column 8, while the observed percentages of larvae killed and migrating per grain appear in column 7.

An inspection of Fig. 2 shows, as before, the close agreement between observed and calculated percentages. The goodness of fit between observed and calculated values of p_t (not shown here) was tested by calculating the value of χ², and there proved to be no significant difference between them. It is therefore reasonable to believe that the elimination of Sitotroga first instar larvae also takes place as a result of random encounters within the grain As before, the proportion of larvae killed as a result of encounters increases with increasing density, while the proportion migrating decreases (column 9). The rate of migration and death of Sitotroga is significantly higher than that of Rhizopertha (χ² corresponds to p<0·01).

The above experiment was now repeated except that first instar larvae of both Rhizopertha and Sitotroga were introduced into the grains together in equal numbers as shown in Table 4. The values for Sitotroga are consistently higher than those for Rhizopertha, although the difference between them is not statistically significant. The expected values of p_i/i were calculated from the values of k obtained from Tables 2 and 3, respectively, and are shown in column 8 of Table 4. An inspection of Figs. 1 and 2 shows that, as before, there is a close agreement between observed and calculated percentages. The calculation of χ² showed also that there is no significant difference between observed and calculated values of p_i for either species (not shown here). The values of p_i/i are (for both species) closely similar whether they are competing intraspecifically or interspecifically, which shows that the degree to which the larvae of either species are affected depends upon that species rather than upon the competing species. The Rhizopertha appear to be slightly the more successful in competition. It should be mentioned that the larvae of either species will attack those of the other. In all the grains dissected in the experiments described below, whenever two larvae were found in the same tunnel at least one of them was always dead.

By the usual method four freshly hatched Rhizopertha larvae were introduced into each of number of wheat grains. Five fresh grains (marked) were then placed in each of the cells containing one infested grain, and the whole returned to the incubator. The fresh grains were dissected, being replaced by an equal number of new fresh grains, once every 4 days, until adults had ceased emerging from the infested grains. The number of larvae migrating to the fresh grains over the whole developmental period was thus observed, instars being checked by measuring the width of the head capsule. Most of the adults had emerged by the 40th day but a few emerged later. The infested grains were all dissected on the last day of observation. The results are shown in Table 5. Most of the migration took place in the first and second instars, and migration had practically ceased by the 20th day. By this time most of the larvae would be entering the prepupal stage. Of the original four larvae per grain the average number which survived to become adults was only 1·14 or 28·5%, 26·1% Of the larvae migrating and 45·4% being killed or dying. Some of the latter, in various instars including one dead, deformed adult, were found on dissecting the infested grains at the end of the experiment.

This experiment was now repeated with Sitotroga. The results (Table 6) here are similar to those just described. Instars were checked as before by measuring the width of the head capsule. Most of the adults had emerged by day 36, but some appeared after longer periods. Migration took place chiefly in the first and second instars and ceased by day 20. The larvae of this species would be in the fourth instar and perhaps beginning to pupate by this time. Of the original four larvae per grain an average number of 1·27 per grain or 31·6% survived to emerge as adults, 22·5 % of the larvae migrating and 45·9 % dying or being killed. Some of the latter, in various instars including one dead deformed adult and one dead pupa, were found on dissecting the infested grains at the end of the experiment.

A third experiment was now carried out which was a repetition of the preceding two except that of the initial four freshly hatched larvae per grain two were Rhizopertha and two Sitotroga. The results are given in Table 7. As before migration occurred chiefly in the first and second instars and had ceased by day 20 of the original four larvae per grain the average number which survived to become adults was 1·23 per grain or 30·8%. Of these survivors 48·7% were Rhizopertha and 51·3% Sitotroga. The proportion of the original larvae migrating was 25·8% of which 45% were Rhizopertha and 55% Sitotroga ; while the proportion dying or being killed was 43·3 % of the original number, of which 54% were Rhizopertha and 46% Sitotroga. As before some of the latter, in various instars including one dead pupa of each species, were found on dissecting the infested grains at the end of the experiment.

What has been observed here is what would happen in a culture in which a large number of wheat grains were present, such as in experimental populations growing in wheat media (unpublished) or in stores of wheat in which these insects are pests. The experiments have provided data about the amount of reduction in numbers for which each of the two methods, migration and death, is responsible, the instars during which migration occurs, the number of larvae capable of successfully completing their development in one grain, and the outcome of competition between Rhizopertha and Sitotroga.

Increasing numbers of freshly hatched larvae were introduced into different wheat grains and each wheat grain placed in a small vial. The whole was then placed in the incubator. There were no fresh grains, each of the vials containing one infested grain only. The vials were examined at regular intervals and the numbers and dates of emergence of adults recorded. The adults were sexed and weighed. We thus have records of the total number of adults emerging, the number emerging per grain, the developmental period, sex ratio and weight. In Table 8 are given the results of experiments on the effect of larval density upon the survival of (a) Rhizopertha and (b) Sitotroga separately. In both cases when there was initially one larva per grain only 80% of the larvae, or 0·8 per grain, survived to become adults. Now in the experiments carried out to investigate whether migration would occur with initially only one larva per grain (vide supra), with both species four of the corpses found in the grains from which no adults emerged were first instar larvae. If these had survived 46 out of 50 or 92 % of the Rhizopertha larvae, and 47 out of 50 or 94% of the Sitotroga larvae would have become adults. When the larvae were introduced into the grains in the second instar, with initially one larva per grain 94 % of the Rhizopertha and 96 % of the Sitotroga larvae survived to become adults (Table 8). Therefore there is a certain mortality in the first instar which is independent of larval encounters.

The majority of the adults of both species emerged within a few days of each other. In Rhizopertha over 95 % emerged between 28 and 38 days after hatching, the mean developmental period of the insects emerging within these limits being 32 days. Over 86 % of the Sitotroga emerged between 26 and 36 days after hatching, the mean developmental period of the insects emerging between these limits being 29 days. The term ‘developmental period’ refers here to the time elapsmg between the hatching of the egg and the emergence of the adult from the pupa. This can be accurately determined for Sitotroga since in the act of emerging from the pupa the adult also escapes from the wheat grain in which the pupa is lying. But owing to the habit of remaining inside the grain for a few days after emergence from the pupa while the adult cuticle hardens, the precise moment of emergence from the pupa cannot be accurately determined for Rhizopertha which have pupated inside wheat grains. What is observed is the moment when the adult eats its way out of the grain, and consequently the values for developmental period are too high. From comparison with data obtained with insects developing in flour, where the precise moment of emergence from the pupa can be observed, the period between emergence and escaping from the grains is approximately 3 days, which makes the actual mean developmental period of Rhizopertha approximately 29 days, or equal to that of Sitotroga. The insects not emerging within the limits of the majority of each species as defined above, had widely divergent developmental periods, always longer than those of the majority. The longest developmental period observed for Rhizopertha was 90 days, and for Sitotroga 120 days. As the developmental periods of these minorities were widely scattered, they were disregarded in calculating the mean developmental periods at each density. The latter did not vary significantly at any density from the mean values given. Now in column 7 b are given the percentages of the surviving Sitotroga whose developmental periods were more than 21 days longer than the average. That is, these insects emerged after the 50th day from hatching. The correlation coefficient between the proportion of insects with such retarded development (column 7 6) and the average number of adults emerging per grain (column 5 b) was calculated, and proved to be significantly different from zero at the 1 % level, and positive. The two sets of variables are therefore highly positively correlated. On the other hand, with Rhizopertha the number of survivors emerging more than 21 days after the average (i.e. after day 53) was never greater than two at any density, and no calculations can be based on such small numbers. The reason for choosing the period of 21 days will appear later. With Sitotroga, in the majority of the grains from which more than one adult emerged, one of the insects had an average rate of development while the rate of development of the other was retarded. The result was that the two insects which developed successfully had widely different rates of development. In this experiment there were actually 56 grains from which more than one Sitotroga adult emerged. The dates of emergence of the adults which had developed in one grain were separated by more than 21 days in 27 = 48·2% of these grains, by 14-21 days in 16 = 28·6 %, and by less than 14 days in 13 = 23·2%. In the grains in which only one insect completed its development the adults emerged within 14 days of each other in 390 = 90% out of 434 grains. In Sitotroga, therefore, if two larvae which have entered the same grain have widely different developmental periods it will be more probable that both of them will successfully complete their development, than it will be if their developmental periods are nearly equal. This is possibly because larvae of approximately the same instar are less tolerant of each other than are those of widely different instars (cf. Flanders, 1933). One larva with a retarded rate of development in a grain containing several normal larvae would therefore have a greater chance of survival than any of the latter. In column 7 b of Table 8 there is a general tendency for the proportion of retarded larvae among the survivors to increase with increasing density, at least up to ten larvae per grain. When there was initially only one larva per grain 3·9% of the survivors had retarded rates of development, which shows that the phenomenon is at least to this extent independent of crowding. Now if it is assumed that a fixed percentage (e.g. 3-9%) of the larvae hatching have constitutionally retarded rates of development then, as the initial number of larvae per grain increases, the absolute number of retarded larvae will increase also. If the survival rate of the latter is greater than that of normal larvae then, since the number of larvae which survive per grain is limited, the proportion of retarded larvae among the survivors will increase. It is also possible, of course, that crowding is a cause of the retarded rate of development (although, as already stated, not always) and that this is the explanation of the results in column 7 b. Above ten larvae per grain perhaps the probability of elimination early in development is so great that the possession of an atypical rate of development is no longer an advantage.

There is no similar phenomenon in Rhizopertha. The dates of emergence of the adults emerging from one grain were separated by less than 7 days in 48 = 78% of the 62 grains from which more than one adult emerged, by 7-14 days in 9=14%, by 14-21 days in 2 = 3·2%, and by over 21 days in 3=4·8%. In the grains in which only one insect completed its development the adults emerged within 14 days of each other in 470 = 96-7% out of 486 grams. Furthermore, the proportion of retarded larvae was always very small.

Overcrowding in the larval instars seems to have had no other effect upon these insects than to increase mortality and to affect the numbers of larvae with retarded development in the way described above. The average developmental period of the majority of the larvae remained unaffected. The sex ratio of the emerging adults did not differ significantly from unity at any density for either species. The average weights of these adults at all densities did not differ significantly from the following values : Rhizopertha males, 1·17 mg. ; females, 1·3 mg. ; Sitotroga males, 1·8 mg.; females, 3·7 mg. The rates of oviposition in wheat of ten of the females which had developed at each density were measured at a density of two grains per female (Crombie, 1942, Tables 1 and 11). The average rates of oviposition for all these females were 9·8 eggs per female per day for Rhizopertha and 112 eggs per female per 5 days for Sitotroga. The rates of oviposition of each species at any density did not differ significantly from these values, respectively. The 143 grains from which one Rhizopertha emerged weighed 7·15 g. (50 mg. per grain) before and 5·50 g. after the insects had developed in them. The faeces and frass were removed from the still uneaten part of the grain before the second weighing. The loss in weight incurred through the development of one Rhizopertha per grain was therefore 1·65 or 23% of the original weight. The average amount of food eaten per larva during development (i.e. the average loss per grain) was 11·5 mg. This figure will be too high since some feeding would have been done by the adults before they ate their way out of the grains after emergence. The loss in weight of the 104 grains (5·2 g.) from which one Sitotroga emerged was 1·92 g. or 37% of the original weight. The average amount of food eaten per larva during development was therefore 18·5 mg. The average numbers of insects surviving per grain when i> 1 (1·16 for Rhizopertha and 1·1 for Sitotroga) are therefore less than the food present in a grain could support if unhindered by other forms of competition between the larvae (vide infra).

The first experiment was a repetition of those just described except that larvae of the two species were present at the same time in equal numbers, i.e. an equal number of each species was competing for the same grain in each vial. The results are shown in Table 9 and Fig. 3.

The total number of insects emerging per grain (column 5, instar I) and the percentage of the total number of larvae surviving to become adults (column 7, instar I), respectively, do not differ significantly from the values of the same variables when çither species was developing alone in the grains (Table 8). This means that the average value of a (equation (2)) here is really equal to those found in Table 8. There is no regular change in the value of a with larval density. The observed values of 100(1 – P) shown in column 7, total, were compared with the values calculated with a=1·21 (not shown), and χ² revealed no significant difference between them. When each species is considered separately the average values of a are 0·69 for Rhizopertha and 0·52 for Sitotroga. The expected values of too (1 – P) were calculated with these values of a, respectively, and proved to be not significantly different from the observed values for each species, respectively, when tested by means of χ² This means that each species reduces the probability of survival of the other in direct proportion to its own numbers, and that there is at all densities a constant relationship between the numbers of the two species which survive. Rhizopertha was always more successful in competition than Sitotroga, the average ratio between them being Rhizopertha to-Sitotroga as 57 : 43 =1·3:1. (The value of χ² calculated in order to compare the numbers of adults of the two species in column 4 (instar I) corresponded to p < 0·01, showing that the difference between the two sets of values is significant.) At larval densities above ten per grain this relationship may seem to change to the disadvantage of Sitotroga, but the agreement of observed and expected values of 100 (1 – P) argues against this, and furthermore with such small numbers the ratio between the survivors of the two species is very sensitive to random fluctuations. Approximately the same relationship is found when the larvae of both species were introduced in the second instar. The weight and sex ratio of the adults were, as before, unaffected by density, and were not significantly different from the values already given with each species developing in the grams separately.

The effect of overcrowding upon the developmental period of the survivors was similar to that already described for each species separately. The percentages of the surviving insects with retarded development are shown in column 8. As before the number of adults emerging per grain (column 5, total) and the proportion of retarded Sitotroga (column 8) are highly positively correlated, the correlation coefficient being significantly different from zero at the 1 % level, and positive. The proportion of retarded Rhizopertha was always very low, comprising only 2 % of the total survivors, while that of retarded Sitotroga comprised 29-5 % of them. The latter proportion is significantly higher than the proportion of retarded insects (3-9 %) when there was initially only one Sitotroga per grain (Table 8 of the (χ² corresponds to p < 0·01). Thus a large proportion of the Sitotroga which survive after competition for the same grain with Rhizopertha had retarded developmental periods. There were fifty-two grains in all from which individuals of both species emerged. The dates of emergence of the individuals of the two species developing in the same grain were separated by over 21 days in 15 = 29% of these grains, by 14-21 days in 26 = 50 %, and by less than 14 days in 11 = 21%. The developmental periods of all the adults were within 10 days of the same length in 300 ( = 90 %) of the 333 grains from which one individual of either species emerged. There were 41 (79%) grains in which the developmental periods of the two species differed by more than 14 days. In 39 (i.e. 75 % of the total 52 grains) of these the rate of development of the Sitotroga was retarded while that Rhizopertha was of the average value (28-38 days), and in only two ( = 3·85%) did the opposite relationship hold. This suggests that the competition of Rhizopertha with Sitotroga larvae whose rates of development differ widely from their own is less severe than with those Sitotroga whose rates of development approximate their own. The proportion of retarded insects among the survivors of Sitotroga increases with increasing density up to ten larvae per grain, when the value of 44% is reached. As before there is a fall in this value with initially more than ten larvae per grain. It is clear that the possession of a proportion of larvae with atypical rates of development is of considerable survival value for Sitotroga, enabling more individuals of this species to survive in the limited space of one wheat gram, both when in competition with Rhizopertha and when only Sitotroga larvae are competing for the grains. The cause of retarded development in Sito-troga larvae is unknown.

The effect of non-contemporaneous entry into grains upon the competition of Rhizopertha and Sitotroga. The initial larval density was kept constant at four first instar larvae per grain (two of each species), but in one series of vials Rhizopertha was introduced 7, 14, 21 and 28 days before Sitotroga (Table io a), while in another series Sitotroga was introduced 7, 14, 21 and 28 days before Rhizopertha (Table 10 b). The results are shown in Table 10 and Fig. 4. Statistically significant differences were discovered by calculating the values of χ² or t, and on the basis of these the following statements are made (in each case p<0·05). (a) The relationship between the two species as observed by their survival (columns 5 and 7), relative to their relationship when both species were introduced at the same time (control), remained always to a greater or lesser extent in favour of the species introduced first (cf. Smith, 1912; Picard, 1922). (b) For the second species the most unfavourable interval at which to follow the first was 7-14 days, (c) As in previous experiments, however, the Rhizopertha were relatively more successful than the Sitotroga. For instance, while the survival of Rhizopertha was unaffected by Sitotroga which follow it by 14 or more days (column 5, and Table 8 a), Sitotroga always had a lower rate of survival when Rhizopertha followed it at any interval of time than when it was alone (column 5 b, and Table 8,6). (d) As the time interval between the introduction of the two species became longer (above 7 days), the number of adults emerging per grain of the first species eventually reached a fairly constant value, while that of the second species gradually increased (column 5). The total number of insects emerging per grain thus increased also. This was achieved by an increase in the number of wheat grains from which more than one adult emerged (column 6), which suggests that the mortality which occurred in the immature stages of the competing insects was not primarily due to lack of food. Additional evidence in support of this conclusion comes from two sources. The first is the observation that up to three and four normal-sized adults of either or both species have been known to emerge from a single grain. Secondly, as already mentioned, the average loss in weight of the wheat grains as a result of the successful development of one insect per grain is only 23 % of the original weight of the grain for Rhizopertha and 37 % for Sitotroga. Now the species introduced first eats food, uses up oxygen and conditions the inside of the grain with its faeces and other waste products. However, if these were the principal causes of the results one would expect that the longer the interval between the introduction of the two species the greater would be the depression of the species introduced second. This was not so, so that the principal cause of the results must have been the active competition of the larvae inside the grains for space (cf. Pemberton & Willard, 1918 ; Lloyd, 1940). On the other hand, 28 days after introduction the first species would usually be in the pupal instar. It is therefore not likely to have engaged in active competition. Now the food provided by one grain is as already mentioned adequate for the complete development of at least two larvae. The reason why the second species survived better in an otherwise uninhabited grain (Table 8) than in a grain in which the first species was preceding it by 21 days or more, is therefore likely to be that the first species conditioned the inside of the grain into which the second species enters. A considerable amount of faeces and frass was always made by the larvae (vide infra), (e) Sex ratio and weight of neither species was affected by competition.

Now in all except two of the 98 grains (column 6) from which both species emerged the interval of time between the emergence of the individuals of each species was greater than 21 days. As already mentioned, as the interval of time between their introduction becomes longer (above 7 days), there is a statistically significant increase in the number of grains from which both species emerged (column 6). Since the rates of development of the majority of individuals of both species would be approximately equal, the members of each species would be in more widely different instars as the interval of time separating their introduction increases. The results therefore agree with those obtained in previous experiments (Tables 8, 9), in which it was found that the mortality of Sitotroga and its competitors which were developing in the same wheat grain was greater among insects with approximately contemporaneous development than in those whose developmental periods differed by a great enough margin (21 days).

Why should the species introduced first have a relative advantage over the other species, and why should it be more unfavourable for the second species to be introduced 7–14 days after the first than either at the same moment or after longer intervals of time ? The following experiment was performed in order to answer these questions. Two larvae of each species were introduced at the same time into each grain, but in each experiment the larvae of one species were in a more advanced instar than those of the other, as shown-in Table 10 c. The numbers involved are rather small, but the results are remarkably similar to those shown in Table 10 a, b, the initially more mature larvae here bearing the same relationship to their competitors as those introduced first in the previous experiment bear to theirs. The mortality of the insects introduced in the first instar was in both species greater when the other competing species was introduced in the second instar than when the latter was introduced in the first or third instar. Now in both species the second instar occupies the period from about 6-12 days after hatching, so that in the previous experiments the larvae introduced first by 7-14 days would be in the second or third instar when the second species was introduced. Those introduced first by 21 days would be in the third or fourth instar by the time the second species was introduced. Furthermore, both the total number of insects emerging per grain, and the number of grains from which both species emerged, were greater when one species was introduced in the first instar and the other the third instar, than when one was introduced in the first instar and the other in the second instar, or both in the first instar. It seems reasonable to suggest that these differences in the severity of competition between larvae of different ages may be the explanation of the greater survival of larvae whose development was not contemporaneous with that of their competitors than of those in which it was (Tables 8, 9). As the larvae enter the grains after increasingly different periods (Table 10) the time during which they are in competition of course becomes shorter, but this does not explain why one species should suffer more when introduced days after, than at the same time as, the other. If the latter result is correct, it must be concluded then that the toleration of the larvae for each other at any moment varies with their relative ages. There is, however, another possible explanation. As mentioned above, Rhizopertha adults remain inside the grain for a few days after emergence before eating their way out. It is possible that these adults may damage other insects in the same grain.. If the latter were following them by only a short interval they would be more likely to be in the pupal stage when these adults emerged than in the larval stage, and perhaps pupae are more susceptible to such damage than larvae. On several occasions dead and partly eaten pupae were found in grains from which other insects had previously emerged as adults, but the data were too meagre to be able to draw any definite conclusions. This explanation would not of course apply to the results obtained when Sitotroga was the first species introduced into the grains.

Similarly, the probability that one Sitotroga will survive is ⁠. The relative probability of survival is therefore Rhizopertha to Sitotroga as 1·3 r : s. The agreement between expected and observed values was tested by calculating the value of χ², and there proved to be no significant difference between them.

There were only 18 grains from which both species emerged in this experiment; in n of these the Sitotroga emerged more than 21 days after the Rhizopertha. As before the weight and sex ratio of the insects remained unaffected by crowding.

I should like to thank Dr A. D. Imms, F.R.S., and Dr W. H. Thorpe for their interest in these experiments and Dr A. R. Miller for his mathematical advice. This work was done while holding a research scholarship from the University of Melbourne.