A Gregarizing Factor Present In The Egg Pod Foam Of The Desert Locust Schistocerca Gregaria

Locusts have a pronounced ability to change their behaviour, development, physiology, morphology and colour in response to changes in population density (Uvarov, 1966). Under crowded conditions, the insects exist as the phase gregaria, whilst when isolated the phase solitaria predominates (Uvarov, 1966). Between these two forms, there exists a continuum, and it is recognised that the transition between the two extremes takes a number of generations. Implicit in this density-dependent transition is the notion that phase characteristics accrue across generations, although the means by which such information is transmitted to the offspring has up to now remained unknown (for a review, see Pener, 1991).

Using a bioassay to quantify the behavioural phase state of individual hatchlings (Roessingh et al. 1993; Roessingh and Simpson, 1994), it has recently been shown that the population density of adult female S. gregaria during mating and oviposition influences the behavioural phase and coloration of the hatchlings from egg pods laid by the females (Islam et al. 1994a). In particular, crowding during oviposition causes females previously reared in isolation to produce offspring with gregarious behaviour. Likewise, isolation of crowd-reared females at oviposition results in the production of offspring that behave more solitariously (Islam et al. 1994b). In a further detailed study, it has been shown that experience of crowding by solitary-reared females for a 2-day period prior to oviposition affects the phase characteristics of the offspring (Bouaïchi et al. 1995). The strength of the effect is dependent on the age at which the adults experience the crowding, increasing in a graded manner with adult age and hence with the proximity of crowding to oviposition. Thus, parents crowded at a late stage in the reproductive cycle produce hatchlings indistinguishable from those of crowd-reared adults. All these observations are consistent with the premise that female S. gregaria, through their previous experience of crowding, are effectively predicting the probability that their offspring will emerge into a high-density population and thus predisposing the behaviour of their hatchlings accordingly.

Grouped oviposition is characteristic of gregarious-phase locusts (Popov, 1958; Stower et al. 1958; Norris, 1963, 1970) and is assisted by interactions between adults which may involve tactile (Ellis, 1959, 1962), visual (Ellis and Pearce, 1962) and olfactory (Nolte, 1963, 1974; Gillett, 1968) stimuli. In the field, oviposition sites of S. gregaria containing egg pod densities of 200–800 m⁻² have been recorded (Stower et al. 1958; Roffey and Popov, 1968). Such behaviour ensures that hatchling locusts are in close proximity to each other, thus promoting and maintaining shifts towards the extreme gregarious phase (Uvarov, 1977). The degree to which the gregarizing effects of crowding at oviposition observed by Islam et al. (1994a,b) and Bouaïchi et al. (1995) are the result of adult crowding itself or are due to high egg pod density has not been previously established and is examined in the present study. Neighbouring egg pods could influence each other by diffusion through the soil of a factor that influences development. Such a phenomenon might explain the effects of crowding at oviposition, and in the present study we test the hypothesis that soil previously used by gregarious ovipositing females is a source of gregarizing factor. The evidence presented here shows that soil is unlikely to be a source of a gregarizing factor.

The question then arises as to how embryonic development is affected in such a way that the behaviour and colour of the offspring vary according to the density experienced by the females at mating and oviposition. One obvious possible explanation is that the female locust produces a causal factor(s) that influences the phase status of hatchlings and which presumably acts by regulating embryonic gene expression. Such a factor could derive from the reproductive tract of the female and affect the subsequent development in the egg at any time from ovulation through to embryonic development.

The accessory glands of the female produce a foam-like material that surrounds the eggs (Szopa, 1981a,b, 1982) and prevents desiccation (Uvarov, 1966). This foam also forms a plug which fills the space above the eggs in the egg pod and provides an escape route for the hatchling locusts. By virtue of its derivation from the female reproductive tract at the time of oviposition and its intimate contact with the eggs, we deduced that egg pod foam could provide an ideal vehicle for the exposure of eggs to any gregarizing factor, and we have tested this hypothesis in a series of experiments described in the present paper. The results of exposing eggs from solitary-reared females and washed eggs from crowd-reared females to foam plugs and foam plug extracts provide the first evidence that egg pod foam of S. gregaria contains a gregarizing factor that influences the development of locust eggs and leads to the production of hatchlings with characteristics of the gregarious phase.

Schistocerca gregaria were maintained under crowded or isolated conditions as described by Roessingh et al. (1993). As in previous publications from our laboratories, the locusts in these two cultures are referred to as crowd-reared and solitary-reared, respectively. These designations generally correspond to the terms gregarious and solitarious, although extremes of the solitarious phase are difficult to produce in the laboratory. In early experiments, fourth-generation solitary-reared females were allowed to oviposit in sand-filled oviposition tubes under different conditions. Each solitary-reared female was mated at least twice, and egg pods were collected throughout its oviposition period. This ensured that sufficient hatchlings were available for colour scoring and behavioural assays. Crowd-reared parents were used to produce egg tubes containing multiple egg pods. For some treatments, solitary-reared locusts were crowded together in groups of four in cages with a 10 cm×10 cm floor. In later experiments, crowd-reared females were used to provide single or multiple egg pods, whilst fourth-generation solitary-reared females were used to provide single egg pods. To determine the effects on the subsequent phase characteristics of the offspring of crowding, isolation and foam plug extracts, the eggs from these pods were incubated under various experimental conditions. Egg pods were collected and incubated for 12–13 days at 30 °C and 60–70 % relative humidity. Those from solitary-reared females were incubated in a solitary locust rearing room, and those from crowd-reared females were incubated in a crowded locust rearing room.

The behavioural assay designed by Roessingh et al. (1993) was used to investigate the responses of individual hatchlings to a stimulus consisting of 50–100 first-instar nymphs from crowd-reared parents in a rectangular arena (33.5 cm×15 cm×10 cm) as extensively described by Islam et al. (1994a,b) and Bouaïchi et al. (1995). In brief, larvae were tested on the first day after eclosion. Stimulus chambers, separated from the central part of the arena by clear, perforated Perspex screens, were situated at each end of the arena. One contained 50–100 first-instar nymphs from crowd-reared parents, and the other was empty. Both chambers were illuminated equally and more brightly than the central area of the arena by using two 8 W fluorescent tubes. The test insect entered the arena through a central hole (2 cm in diameter) in the floor. Typically, hatchlings with behavioural characteristics typical of the gregarious (i.e. crowd-reared) phase moved towards the stimulus group of hatchlings, although a range of behavioural responses was recorded using a lap-top computer. The data were analysed using logistic regression to reduce the multivariate data to a single variable (‘behavioural phase state’). The behavioural elements retained in the logistic regression model included x-distance, track straightness, track speed, distance moved, angle per turn, crouch frequency, grooming-time frequency, grooming frequency, jumping frequency, walking frequency, walking-time frequency and leg movement frequency (Roessingh et al. 1993). For each set of experiments, the model was constructed using crowd-reared and solitary-reared individuals. The resultant predicted probabilities for the behavioural phase state of the insects were used as the response variable in subsequent analyses testing the effects of the experimental treatments. Values for probability (isolated) were ranked, and analyses of variance (ANOVAs) undertaken on the normalised ranks (see Islam et al. 1994b; Bouaïchi et al. 1995). Frequency histograms of values for the behavioural phase state of individual hatchlings (probability isolated) ranging along the x-axis from 0 to 1 in bins of 0.1, as predicted by logistic regression, have been used to demonstrate the effects of experimental treatments. Control insects from the crowd-reared culture lie very largely in the 0–0.01 bin and those from the solitary-reared culture lie very largely in the 0.9–1.0 bin, reflecting the relative probabilities of insects from these two cultures having the behaviour typical of the solitary (i.e. solitary-reared) phase.

Assessment of hatchling colour on the day of hatching was made using a five-point scale as described by Islam et al. (1994a,b), where 1 corresponded to hatchlings with a uniform green ground colour with no black markings and 5 corresponded to hatchlings in which the ground colour is entirely obscured by heavy black markings over more than 80 % of the body surface. These two extremes correspond to the coloration of hatchlings from the solitary-reared (i.e. solitarious) and crowd-reared (i.e. gregarious) cultures, respectively. Analyses of variance using normal scores of the ranks for colour as the response variable were undertaken (Islam et al. 1994a,b; Bouaïchi et al. 1995).

A range of observations and experimental manipulations was performed as summarised in Table 1. To maintain clarity, descriptions of each experiment are incorporated in the appropriate Results section.

Since the density of females at oviposition appeared to be a key factor influencing phase characteristics such as behaviour and coloration in S. gregaria offspring (Islam et al. 1994b), an experiment using solitary-reared females was designed to distinguish whether this was due to the degree of crowding of the females at the time of oviposition or to the resulting density of egg pods.

Four treatments were devised as summarised in Table 1. In the control treatment (treatment 1), single egg pods laid by isolated, solitary-reared females were collected. The behavioural phase state and coloration of a total of 53 hatchlings from four egg pods was assayed. For treatment 2, in which solitary-reared females, isolated at oviposition, laid egg pods into tubes already containing egg pods, it was impossible in practice to collect several freshly laid egg pods from solitary-reared insects in a single oviposition tube. Instead, five egg pods from solitary-reared females were each collected in individual tubes each containing a minimum of at least three freshly laid egg pods from crowd-reared females. Fifty-two hatchlings were assayed. For treatment 3, using solitary-reared females crowded at oviposition and laying a single egg pod, solitary-reared females were paired, as usual, with solitary-reared males for 24 h in individual cages. After mating, the females were isolated for 2 days prior to oviposition. For the period of oviposition, the normally solitary females were crowded together in two cages, each containing four females. The females were allowed to oviposit only a single egg pod per tube. Forty-six hatchlings from three egg pods were assayed. For the final treatment (treatment 4), solitary-reared females were crowded at oviposition and allowed to lay three or more egg pods per oviposition tube within 2–3 days. For these assays, 47 hatchlings from five egg pods were assayed. After collection, all the egg pods were incubated at 30 °C for 12–13 days, and then separated in individual containers in which they hatched. All the hatchlings were reared in isolation and fed until used for behavioural assays on day 1.

Crowding at oviposition, at a density of four solitary-reared females per cage (treatments 3 and 4), was found to have a significant gregarizing effect [higher frequency of individuals with low probability (isolated) values] on the behaviour of subsequent progeny, in contrast to the lack of gregarizing effects [higher frequency of individuals with high probability (isolated) values] resulting from oviposition by a single solitary-reared female per cage (treatments 1 and 2) (Fig. 1). Analysis of variance of the probability (isolated) values showed that the differences in the behaviour of the hatchlings due to the density of ovipositing females were highly significant (P<0.001). The behaviour of hatchlings that originated from pods laid by solitary-reared females in a tube containing three or more other egg pods (treatments 2 and 4) was found to differ significantly (P=0.001) from that of hatchlings derived from single egg pods laid by solitary-reared females either in isolation (treatment 1) or in a crowd (treatment 3) (Fig. 1). As discussed below, our inability to use solitary-reared egg pods to provide multiple egg pod conditions may have confounded these results, but there is no evidence that this was the case in a comparison of treatments 1 and 2, the most sensitive comparison. There was no significant interaction between female density and egg pod density (P=0.071). The frequency distributions of the hatchlings produced in treatments 1–4 showed that, at densities of more than three pods per oviposition tube, the behavioural phase state of the resultant hatchlings was significantly shifted towards gregarization in contrast to that of the hatchlings from a single egg pod per tube.

The colour of hatchlings from egg pods at densities of three or more per oviposition tube (treatments 2 and 4) was found to be significantly darker than that of hatchlings from a single pod per tube (treatments 1 and 3) (Fig. 2) (P<0.001). In contrast, female density at oviposition did not affect hatchling colour (P=0.413), nor was there any interaction between female and egg pod density (P=0.226). Thus, a high density of egg pods, although from crowd-reared females, rather than simply a high density of females was found to be more effective at promoting gregarious pigmentation of the hatchlings.

A significant correlation was found to exist between behaviour and coloration of hatchlings (Spearman’s rank-order correlation coefficient r_s=0.4925; P<0.0001), indicating that the darker, black hatchlings behaved more gregariously than their lighter, green counterparts.

In summary, both crowding of solitary-reared females at the time of oviposition and high egg pod densities promoted behavioural gregarization, although the former appeared to be a rather more potent factor. Egg pod density alone appeared to promote the development of hatchlings with dark patterns characteristic of the gregarious phase.

Because egg pod density appeared to influence some phase characteristics, we hypothesised that there might be a chemical factor in egg pods that influences surrounding egg pods and that sand in which locusts have previously laid egg pods could conceivably contain this factor(s). This might, in turn, influence the phase characteristics of hatchlings from egg pods subsequently laid into the used sand. Accordingly, we tested this idea using ovipositing solitary-reared females and sand previously used for oviposition.

For the control treatment (treatment 5), unused clean silver sand was repeatedly washed in hot tap water and then oven-dried at 140 °C. Control solitary-reared females were offered this sand as an oviposition medium for laying single egg pods. The colour scores and behaviour of 100 hatchlings from 10 egg pods were assayed. Sand used 24 h previously for oviposition by solitary-reared females was sieved to remove any faeces, foam plug or insect fragments, but the sand was not washed. The sand was then used as a medium for the oviposition of single egg pods from solitary-reared females (treatment 6). Fifty-three hatchlings from four egg pods were assayed. In the final treatment (treatment 7), sand used 24 h previously in oviposition tubes from the crowd-reared locust culture was offered as an oviposition medium to the solitary-reared females. Forty-two hatchlings from four egg pods were assayed.

The behavioural phase characteristics and coloration of hatchlings were unaffected when sand used 24 h previously for oviposition by either crowd-reared or solitary-reared locusts was used to collect further egg pods from solitary-reared females. Hatchlings derived from egg-pods laid into clean, unused sand (control) did not differ significantly from hatchlings emerging from egg pods laid into sand from the solitary-reared or crowd-reared cultures either in behaviour (Fig. 3) (P=0.6694 and P=0.2004 respectively) or colour (P=0.5144 and P=0.3577 respectively) (Fig. 4).

The experiments described above suggested that crowding at oviposition and a high density of egg pods led to gregarization of the offspring of solitary-reared females. Conversely, it was hypothesised that isolation of crowd-reared females at oviposition or isolation of their egg pods should lead to solitarization. We reasoned that these factors could be manipulated to induce behavioural and/or pigmentary solitarization in crowd-reared locusts or to reverse the phase characteristics in solitary-reared insects. To test this, the series of experiments described below was carried out, as summarised in Table 1. In all these experiments, selection of test eggs was randomised with respect to the position of the eggs in the pods (i.e. top, middle and bottom), particularly since Papillon (1960) reported that the hatchlings from the lower third of the pods of crowded S. gregaria were sometimes small and pale.

Since some factor in the egg pods might influence phase, we reasoned that isolation of eggs from egg pods should enhance solitarization. Single egg pods laid by crowd-reared females were collected in oviposition tubes, and individual eggs from each egg pod were separated at various times after oviposition (treatment 8). Eggs were either washed in saline (treatment 8b) or left unwashed (treatment 8a) before being placed individually in containers on cotton wool moistened with distilled water. The containers were kept sealed and incubated at approximately 30 °C for 12–13 days until hatching. Egg pods from which the eggs were neither separated nor washed in saline were used as a control. In total, 592 and 443 hatchlings emerged from unwashed and washed eggs, respectively, and were used for the colour score assessment.

Eggs which were separated within the first hour of oviposition clearly resulted in a significantly greater proportion of green hatchlings (colour score 1) than any other treatment group (P<0.001) (Fig. 5). Additionally, it was noted that, when similar early-separated eggs, surplus to the requirements of the experiment were crowded, for example at a minimum density of 10–12 individual eggs in a closed container, they did not produce green hatchlings. Although the number of green hatchlings decreased markedly within 1 day after laying, the mean colour score of the hatchlings (4.18) was light enough to be significantly different from that of the hatchlings from the control group (4.66) (P<0.001). The colour of the hatchlings from eggs separated on days 3, 6, 9 and 12 after laying did not differ from those of the control treatment. Hatchlings from eggs washed with saline within an hour of laying, when compared with those from unwashed eggs separated within an hour of laying, were not found to differ significantly in respect of their colour (P=0.1142), suggesting that a very early separation has the same effect as that of washing. Interestingly, a differential solitarizing effect of washing on hatchling colour was found for the eggs separated within a day of laying compared with the control (P<0.001), whilst washing of eggs more than 3 days old had no further solitarizing effect on hatchling colour.

The above experiment, using unwashed eggs, was repeated using times up to 5 h to provide more detail of the temporal effects of early separation on colour development (treatment 9). For this treatment, it was necessary to know the exact time of oviposition and, therefore, egg pods were collected from crowd-reared locusts actually observed ovipositing. The eggs were not washed in saline. Colour scores of a total of 227 hatchlings from 15 egg pods were recorded.

Nearly half of the hatchlings were found to be green when the eggs were separated within 15–20 min of oviposition followed by isolated rearing of the hatchlings (Fig. 6), and the majority of the rest of the hatchlings were lighter in colour than those of the control group. Eggs separated after an hour also had approximately 30 % green hatchlings and a considerable number of lighter hatchlings. Separation after 2 h did not yield any green nymphs, although approximately a quarter of the hatchlings had a colour score of 3. Significantly higher proportions of dark hatchlings (scores from 3 to 5) emerged from eggs separated 3, 4 and 5 h after oviposition (Fig. 6).

In summary, early separation of presumptive crowd-reared eggs from egg pods laid by crowd-reared females led to solitarization of the colour of hatchlings. A factor, either in or around the eggs, removed by early separation clearly promoted gregarization.

The results obtained up to this point suggested that hatchlings with gregarious behaviour and coloration typical of crowd-reared locusts resulted from incubation of multiple, crowded eggs, and that early separation of these eggs reversed this effect. Logically, we considered the possibility that presumptive solitary hatchlings from egg pods freshly laid by solitary-reared females could be gregarized by incubating the latter with freshly laid eggs from crowd-reared females or with the foam plug from egg pods from crowd-reared females.

In the control treatment (treatment 10), 10–20 individual eggs from a freshly laid (less than 8 h old) egg pod from a fourth-generation solitary-reared female were placed around a clump of 10–20 eggs from another freshly laid egg pod from a similar solitary-reared female. The eggs were arranged so that they touched each other, and they were kept in a plastic container on moist cotton wool. The container was kept sealed and incubated at approximately 30 °C for 12–13 days, after which the individual solitary eggs were separated and allowed to hatch in isolation. Of 40 hatchlings from six egg pods scored for colour assessment, 17 from four egg pods were used for the behavioural assay. For the experimental treatment (treatment 11), 10–20 individual eggs from a solitary-reared female were placed around a cluster of 10–20 eggs from a freshly laid (less than 1 h old) pod from a crowd-reared female so that all the eggs touched each other. The eggs from the solitary-reared female could be distinguished from those from the crowd-reared female by their smaller size and paler colour. Twenty-three hatchlings from three egg pods were used for the behavioural assay. In addition, the colour scores of a further 36 hatchlings from four egg pods were also recorded.

When incubated with gregarious eggs (treatment 11), eggs from solitary-reared females gave rise to hatchlings which showed a degree of behavioural gregarization when compared when control eggs from solitary-reared females (treatment 10) (Fig. 7) (P=0.0398). Likewise, darker hatchlings were found when eggs from solitary-reared females were incubated with eggs from crowd-reared females (Fig. 8) (P=0.0103) than were found in the control group.

Egg pods from crowd-reared females were collected in oviposition tubes fitted with vertical partitions dividing the tube into four sections. These loose-fitting partitions were designed to enable collection of intact individual egg pods from tubes used for oviposition by crowd-reared females. Such partitioned tubes facilitated removal of the egg pods without damaging the foam plugs and prevented egg pods being laid closely in such a way that they adhered to each other. A longitudinal slit was made in the isolated foam plug (still soft and shiny in appearance), and 5–6 freshly laid (less than 8 h old) eggs from solitary-reared females were carefully embedded into it. The walls of the slit were then closed. The foam plug containing eggs was then incubated in containers on cotton wool moistened with distilled water. The containers were kept sealed and incubated at approximately 30 °C until hatching. Finally, just prior to hatching, the treated eggs were separated for isolated hatching into individual containers. In total, 18 hatchlings from three egg pods were used for the behavioural assay and a further 13 hatchlings from three egg pods were scored for their coloration.

There was a significant gregarizing effect of incubating eggs from solitary-reared females in the foam from egg pods laid by crowd-reared females (treatment 12) on the behaviour of the hatchlings (P=0.0280). Similarly, when treated with foam from egg pods laid by crowd-reared females, hatchlings from eggs laid by solitary-reared females became considerably darker (P<0.001).

A complementary treatment to that described above was devised in which the phase of the insects was reversed: eggs from crowd-reared females were incubated in foam from egg pods laid by solitary-reared females (treatment 13). Since foam plugs from egg pods laid by solitary-reared females failed to induce green coloration in the hatchlings emerging from the treated crowd-reared eggs, the behavioural assay of the hatchlings was omitted. Only the colour scores of 44 hatchlings from 10 egg pods were assessed.

In contrast to the pigmentary effect of foam from egg pods from crowd-reared females on solitary eggs, the foam plug of egg pods from solitary-reared females (treatment 13) failed to induce a significant solitarizing effect on the colour of hatchlings emerging from treated crowd-reared eggs (P=0.8288).

As part of the series of experiments described above (treatments 10–15), these treatments were used as controls for the separation of eggs and examined the effects on both hatchling behaviour and coloration. Eggs from pods oviposited by crowd-reared females were separated 8–9 h after laying and placed in containers on cotton wool moistened with distilled water (treatment 14). The containers were kept sealed and incubated at approximately 30 °C hatching. Of 88 hatchlings from six egg pods scored for colour, 20 hatchlings from three egg pods were tested in the behavioural assay. In the crowd-reared control treatment, 10–20 freshly laid (less than 1 h old) gregarious eggs were incubated with a cluster of 10–20 eggs from another freshly laid gregarious pod (treatment 15) in a similar manner to that described for treatment 10 above. In total, 22 hatchlings from three egg pods was tested in the behavioural assay in addition to a total of 80 hatchlings from five egg pods scored for their coloration.

Separation of crowd-reared eggs from egg pods within a few hours of oviposition (treatment 14) led to a marked behavioural solitarization of the hatchlings (Fig. 7) (P=0.0275) compared with crowd-reared controls (treatment 15), in which crowd-reared eggs were incubated within a cluster of eggs from other crowd-reared females, and a significant proportion of light-coloured hatchlings (colour scores 2 and 3) were found (P<0.001).

In treatments 10–15, a highly significant correlation was found between the probability (solitary) values and the colour scores of the test hatchlings (Spearman’s rank-order correlation coefficient, r_s=0.5709; P<0.001).

Separation of eggs from crowd-reared females clearly caused hatchlings to emerge with behaviour and coloration typical of solitary insects. Both the eggs and foam plugs of egg pods from crowd-reared females appeared to be a source of gregarizing factor. In contrast, there was no evidence for a solitarization factor in the egg pods of solitary-reared S. gregaria.

The results from the experiments above indicated that foam plugs might contain a gregarizing factor. To investigate this, crude extracts of gregarious foam plugs were obtained as follows. At least 10 foam plugs from freshly laid egg pods from crowd-reared females were collected in partitioned oviposition tubes designed to enable collection of intact individual egg pods. These were carefully cleaned with a paintbrush, to remove the attached sand particles, and cut into small pieces. Samples of foam plug were then homogenised in locust saline, acetone or ethanol (analytical grades) using a ground-glass homogeniser and were then filtered. The homogenates were prepared immediately before use at a stock concentration of one foam plug per 500 μl of solvent. Containers were each provided with a pad of cotton wool moistened with distilled water, and a 4.25 cm filter paper circle (Whatman) with an M-shaped crease in it was treated with concentrations of 0.2, 0.4 or 1.0 foam plug equivalents per filter paper (treatment 16). The acetone and ethanol were allowed to evaporate at room temperature (<2 min) from the filter papers before they were placed in the containers. The saline-treated filter papers were dried in the same way, and the minute amounts of salts remaining from the locust saline were taken up by moisture in the container. Five newly laid (less than 8 h old) eggs from second- and third-generation solitary-reared females were placed on the groove in the filter paper. The treated filter paper and the test eggs were covered with another non-folded, flat filter paper of the same size. The containers were sealed and incubated at 30 °C until eclosion of the eggs. Solvent controls, in which the filter paper was treated with solvent alone, and untreated eggs, in which the filter paper was untreated, were maintained for comparison. On the day of hatching, the nymphs were scored for their coloration.

The saline extract resulted in a considerable number of hatchlings with black patterns characteristic of gregarious phase hatchlings (colour scores 3 and 4), although very black hatchlings (colour score 5) were not produced (Fig. 9). Nevertheless, treatment of eggs with saline extracts of foam plug clearly resulted in a greater proportion of darker hatchlings than treatment with either acetone or ethanol (P<0.001).

Since saline extraction of foam plugs appeared to produce an active factor, a further experiment was conducted to ascertain whether treated eggs would respond in a dose-dependent manner to this extract. Saline extract at concentrations of 0.05, 0.1, 0.5 and 1.0 foam plug equivalents per filter paper was used to treat filter papers on which eggs from solitary-reared females were incubated in sealed containers (treatment 17). Saline controls and untreated eggs were maintained as above, and the colour scores of the hatchlings were recorded.

Dark-coloured hatchlings could be produced only by treating freshly laid (less than 1 day old) solitary eggs with 1.0 foam plug equivalent of the extract, while doses of 0.5 and 0.1 foam plug equivalents produced only lighter hatchlings (Fig. 10). The saline-extractable gregarizing factor in foam plugs thus appeared to act in a dose-dependent manner.

To define the period of susceptibility of eggs from solitary-reared females to the foam plug extract, and also to determine the period for which fresh saline extract would remain effective after preparation, the following 3×3 experimental design was devised (treatment 18). Each of 10–15 individual eggs from a single egg pod from a solitary-reared female was treated with 50 μl of 1.0 foam plug equivalents of the saline extract on day 0 (day of laying). Similar batches of 10–15 eggs from the same pod were treated in a similar way on days 1 and 3. Likewise, freshly prepared extracts applied on the day of preparation (day 0) and those stored at −70 °C for 1 or 3 days were applied to the solitary eggs on days 0, 1 and 3 after oviposition. Eggs were incubated in containers provided with moist cotton wool, as described above, or in containers with 1.2 % agar (FSA Laboratory Supplies, UK) to provide sufficient moisture for the test eggs. Controls for the test eggs and the extract (saline only) were maintained, and the nymphs were scored for their coloration on the day of hatching.

When applied to day 0 or day 1 eggs from solitary-reared females, the freshly prepared extract produced a significantly greater number of dark hatchlings than the controls (P<0.001 and P=0.0036, respectively) (Fig. 11). The number of dark hatchlings produced did not differ between eggs treated with fresh extract on day 3 compared with the controls (P=0.0650). The results indicate that solitary eggs are influenced by the gregarizing factor in foam plug extracts for up to 1 day after oviposition.

When applied on the day of preparation, foam plug extract produced a significantly greater number of dark hatchlings (P<0.001) than 1 day (P=0.2541) or 3 days (P=0.2793) after preparation, when it was found to be ineffective (Fig. 11). These data suggest that, with time, the extract becomes degraded in some way, despite being stored at −70 °C.

In the experiments described above, foam plug extracts were shown to cause gregarization in eggs from solitary-reared females. In addition, washing of gregarious eggs was shown to result in solitarization of offspring. Logically, we hypothesised that treating washed gregarious eggs with foam plug extracts should restore the production of gregarious hatchlings, and the following experiment was conducted. Eggs from a total of four egg pods from the crowd-reared locust culture were washed three times in locust saline within 1 h of oviposition, and the eggs were incubated on moist filter papers in sealed containers (treatment 19). The behavioural phase state of 52 hatchlings was determined using the behavioural assay. A further batch of washed eggs from four egg pods from crowd-reared females was incubated on filter papers treated with locust saline extracts of foam plug at a concentration of 1.0 foam plug equivalent per filter paper (treatment 20). The behavioural phase state of 44 hatchlings was determined using the behavioural assay. The results were compared with those obtained with solitary-reared (treatment 21) and crowd-reared (treatment 22) control insects.

Washing of eggs from crowd-reared females within 1 h of oviposition did not, in this experiment, result in behavioural solitarization in all the offspring, and a significant minority of eggs produced hatchlings typical of the crowd-reared culture (Fig. 12). Accordingly, treatment of washed eggs from crowd-reared females with foam plug extract did not then produce a significant change in the frequency distribution of hatchling types compared with that seen with washed, control eggs, although median values were in the expected range (Fig. 12). It would appear, in contrast to previous experiments, that washing of crowd-reared eggs was not wholly effective at suppressing gregarious behaviour in the hatchlings. As discussed below, either a proportion of the eggs was already committed to the gregarious phase or washing was ineffective for some other reason.

As part of an initial series of experiments to characterise the gregarizing factor in egg pod foam, saline extracts of the foam plugs from six egg pods laid by crowd-reared females were prepared as above. For the control treatment (treatment 23), 36 eggs from three egg pods from solitary-reared females were incubated in closed containers on filter papers treated with this foam plug extract at a concentration of 1.0 foam plug equivalent per filter paper. A similar experimental group of 35 eggs from four egg pods was treated with a portion of the same foam plug extract that had been placed in a water bath at 100 °C for 10 min prior to application to the filter paper (treatment 24). The behaviour of the resultant hatchlings was assayed (Fig. 13).

Heat treatment of the saline extract resulted in a considerable loss of gregarizing activity compared with the non-heat-treated control group (Fig. 13), in which the intact, neat extract caused significant gregarization (P=0.001). These results suggest that, to some degree, the saline extract is heat-sensitive.

In further experiments to characterise the gregarizing factor in egg pod foam, saline extracts of foam plugs were prepared as above and placed in a 3 kDa ultrafiltration device (Amicon) and centrifuged at 10 000 g for 30 min. The filtrate (<3 kDa) was then used to treat eggs from pods laid by solitary-reared females (treatment 25). The retentate was recovered from the ultrafiltration device using a brief spin, and a further batch of similar eggs was treated (treatment 26). In all, 42 hatchlings from five egg pods for the filtrate and 18 hatchlings from two egg pods for the retentate were examined using the behavioural assay.

Hatchlings from eggs treated with the filtrate behaved significantly more gregariously than did eggs treated with the retentate (Fig. 13). The gregarizing activity of the foam plugs of crowd-reared, gregarious locusts thus appears to be due to a small (<3 kDa) hydrophilic factor.

Crowded oviposition is characteristic of gregarious phase S. gregaria (Uvarov, 1966), and from previous studies it was concluded that this was a key factor in determining the phase status of hatchlings in this species (Islam et al. 1994a,b; Bouaïchi et al. 1995). The results presented here confirm that crowding of solitary-reared females at oviposition resulted in a highly significant shift towards behavioural gregarization in the resulting hatchlings (Fig. 1, treatments 3 and 4). Conversely, isolation of crowd-reared females at oviposition resulted in the production of hatchlings which behaved solitariously (Fig. 1, treatments 1 and 2). In the present study, the crowded ovipositing females were kept in a group of four females per rearing cage. This was equivalent to 400 locusts m⁻² and realistically reflects the field situation, where a maximum post-hatching density of 500 locusts m⁻² was recorded (Roffey and Popov, 1968). Nevertheless, it was unclear whether this effect was simply due to the crowding itself or whether the effects on hatchlings were due to the resultant high density of egg pods. Field observations have shown that females of S. gregaria lay their egg pods in dense groups ranging from 200 egg pods m⁻² (Stower et al. 1958) to 800 egg pods m⁻² (Roffey and Popov, 1968). In the laboratory, egg pod densities of 2400–4000 m⁻² were obtained by allowing females to lay 3–5 egg pods in 4 cm diameter oviposition tubes. By manipulating egg pod density it was possible to show that, irrespective of the density of females at oviposition, the density of egg pods also influenced the behaviour (Fig. 1) and colour (Fig. 2) of resulting hatchlings to some degree, although its effects on behaviour (Fig. 1, treatments 2 and 4) were somewhat less potent than those of female crowding (Fig. 1, treatments 3 and 4). The effects on colour may have been influenced to some degree by the use of multiple pods from crowd-reared females (Fig. 2, treatments 2 and 4), but this was clearly not so for the behavioural component (Fig. 1, treatments 2 and 4).

Since egg pod density appeared to affect the phase characteristics of offspring, we argued that this was most likely to occur as a result of diffusion through the soil of some factor from the egg pods. Interestingly, when sand previously used for crowded oviposition by the crowd-reared, gregarious culture was used as a medium for oviposition by test females, there was no effect on the phase characteristics of the hatchlings (Figs 3, 4, treatment 7), although it is possible that the sand had lost some of its potential activity by 24 h after the previous oviposition. It has been reported that females of gravid S. gregaria did not aggregate near sand into which other S. gregaria had previously laid egg pods nor did they lay a significantly increased number of egg pods in comparison with females ovipositing in unused sand (Norris, 1970). Clearly, species differences may exist. In the present study, the failure to demonstrate any factor in used sand causing gregarization may have several causes. First, it suggests that any factor emanating from egg pods may be present only whilst the egg pods are present in the soil and, second, that it may be short-lived in its effect on eggs. As discussed below, subsequent experiments on the gregarizing factor confirm the interpretation that the factor is active for a limited period only and that any attempt to demonstrate a soil-borne factor would require the soil surrounding fresh egg pods to be taken immediately after oviposition.

The hatchlings from isolated egg pods had markedly more solitarious characteristics (Figs 1, 2, treatments 1 and 3) than those from egg pods laid at a high density (Figs 1, 2, treatments 2 and 4). Separating individual eggs from egg pods was a logical extension of this and, if carried out on the day of oviposition, clearly resulted in further solitarization of eggs from crowd-reared parents (Fig. 5, treatment 8; Fig. 6, treatment 9; Figs 7, 8, treatment 14), irrespective of egg pod density. Care was taken in this procedure to remove all traces of foam from the surface of the eggs. Significantly, similar results were obtained irrespective of whether the eggs were simply separated (Fig. 5; treatment 8a) or were washed with saline as well as separated (Fig. 5, treatment 8b). The most likely interpretation of these experiments is that the factor causing gregarization was not present in large quantities on the outside of the eggs but was largely lost when the eggs were separated from the foam of the egg pod.

The evidence that the foam of the egg pod is the source of the factor leading to gregarization is compelling, and the remainder of the experimentation discussed here centres on this premise. Although incubating freshly oviposited, separated eggs from solitary-reared females with eggs from crowd-reared females resulted in a degree of gregarization (Figs 7, 8, treatment 11), considerably more marked behavioural gregarization was obtained by incubating similar separated eggs in fresh foam plugs from egg pods laid by crowd-reared females (Fig. 7, treatment 12). The effects on behavioural phase characteristics (Fig. 7; treatment 12) were especially marked, whilst the effects on colour (Fig. 8, treatment 12) were less so. In contrast, the foam plugs from solitary-reared females had no effect on eggs from crowd-reared females and a high proportion of very black hatchlings was obtained (Fig. 8, treatment 13), signifying that only the egg pods from crowd-reared females possessed this factor. We deduce that females crowded during oviposition produce the factor, whilst solitary-reared females add little or no factor to their egg foam, and this results in behavioural solitarization and lighter-coloured hatchlings.

Evidence that egg pod foam from crowd-reared females contains a gregarizing factor was confirmed in two sets of experiments. Very pronounced behavioural gregarization was observed in hatchlings following incubation of solitarious eggs with foam plugs from pods laid by crowd-reared females. Significant gregarization, shown by the appearance of darker hatchlings, was also obtained when eggs from solitary-reared females were treated with foam plug extract (Figs 9–11, treatments 16–18), although very dark hatchlings with the highest colour score were not obtained. A similar result was also expected when washed eggs from crowd-reared females were treated with foam plug extract (Fig. 12, treatment 20). In the latter case, however, initial washing of eggs from crowd-reared females immediately after oviposition did not significantly increase the likelihood that hatchlings would emerge with behavioural characteristics of the solitarious phase (Fig. 12, treatment 19), as was demonstrated in treatment 14 (Fig. 7). Foam plug extract was thus unable significantly to reverse this. The reason for this discrepancy may lie in the degree of exposure of the eggs in the oviducts to the gregarizing factor before oviposition. Possible changes in the length of time that eggs remain in the oviducts before oviposition could affect their degree of prior commitment to the gregarious phase and may need to be controlled in future studies of this type.

That eggs appear to respond in a dose-dependent manner to the gregarizing factor extracted in saline from the foam plugs of egg pods laid by crowd-reared locusts was shown by applying the extract to separated eggs from solitary-reared females at doses of 0.1–1.0 foam plug equivalents per filter paper (Fig. 10, treatment 17). The highest dose produced the greatest number of dark hatchlings although, as with earlier experiments, very dark hatchlings with high scores were not obtained. Doses of less than 0.1 foam plug equivalent were too dilute to have any effect.

When eggs, which otherwise would have hatched into black gregarious-phase hoppers, were separated from the egg pod immediately after oviposition, the resultant offspring were green (Fig. 8, treatment 14) and behaviourally solitarious (Fig. 7, treatment 14), whilst separation of eggs at a series of later times gave rise to increasing numbers of black, gregarious hatchlings in a time-dependent manner (Figs 5, 6, treatments 8 and 9). Early separation of eggs and removal of foam thus removed the stimulus resulting in gregarization of offspring. Removal of foam later than 24 h after oviposition was without effect and, indeed, separation at 5 h after oviposition (Fig. 6, treatment 9) resulted in most offspring remaining black like the controls (Fig. 8, treatment 15). There are two possible explanations for these results. First, the gregarizing effect of the foam may degrade in some way and, second, the eggs may lose their ability to respond to it. Experiments in which saline extracts of foam plug were used to treat eggs from pods laid by solitary-reared females (Fig. 11, treatment 18) showed both these possibilities to be correct.

When eggs from pods laid by solitary-reared females were treated at various times after oviposition with freshly prepared foam extracts from pods laid by crowd-reared females, the hatchlings had the black coloration and patterning typical of gregarious locusts (Fig. 11, treatment 18). This was only true when the extract was applied within the first 24 h after oviposition (Fig. 11, treatment 18a) and it implies that the eggs are responsive only for a limited period. There are two pieces of evidence which suggest that the gregarizing factor in the foam is effective only for a limited period. First, when kept for periods longer than 24 h, even in a freezer, the extracts lost their ability to alter the phase of eggs from solitary-reared females (Fig. 11, treatment 18), although it is acknowledged that freezing might have inactivated certain types of material. Second, whilst diffusion of some compound was implied by the fact that egg pod density influenced offspring phase (Figs 1, 2, treatments 1–4), there was no effect on the hatchlings of oviposition into old sand previously used for oviposition by the gregarious culture (Fig. 3, treatment 7), as noted above.

The success of the approach taken in this and our previous studies has largely depended on the development of the behavioural assay (Roessingh et al. 1993; Roessingh and Simpson, 1994), but we have also used hatchling colour as a key characteristic of phase. Of the two characteristics, behaviour is by far the more significant in the genesis of locust swarms (Roessingh et al. 1993; Bouaïchi et al. 1996). Although colour and behavioural gregarization were significantly positively correlated (dark, black hatchlings behave gregariously and lighter, green hatchlings behave solitariously), it was evident for solitary-reared females that egg pod density had a more profound effect on the colour of hatchlings than the density of females at oviposition (Fig. 2, treatments 1–4). A similar result was noted previously where crowding of solitary-reared females during oviposition led to significantly more pronounced gregarious behaviour in the offspring but not to significantly darker hatchlings (Islam et al. 1994b). For crowd-reared females, hatchling colour was influenced by both female crowding and egg pod density (Fig. 2, treatments 1–4). These results suggest that, whilst colour and behaviour are frequently found to respond together to density changes, the two phase characters are not necessarily tightly linked.

It has been suggested that a high humidity and isolated conditions for the eggs might, in some way still unknown, make the hatchlings lighter and greener in colour, whereas drier and crowded conditions make them darker and blacker (Hardie and Lees, 1985). Interestingly, the results presented here clearly show that, apart from pigmentary changes, hatchlings emerging from early separated gregarious eggs incubated in isolation had a behavioural phase state significantly shifted towards that of the hatchlings of solitary-reared parents (Fig. 7, treatment 14). Thus, pigmentary changes appeared to be associated with both the amount of foam present around individual eggs and the length of time that the eggs remained together in a pod with the foam plug. Hunter-Jones (1964) recorded the development of separated eggs and noted no change of colour of the hatchlings. This apparent contradiction presumably resulted from the fact that the eggs were incubated as groups of 10 in glass tubes. This would probably have considerably raised the threshold for the development of a dark colour.

A dark-colour-inducing factor in L. migratoria has recently been described by Tanaka and Pener (1994). This factor is reported to be a heat-stable neuropeptide, extractable in methanol or saline. The authors state that, following repeated injections of the extract into the nymphs, they obtained very darkly coloured insects mostly devoid of black patterns. In contrast, in the present study, the gregarization factor led to the development of black patterning characteristic of gregarious nymphs. Simply darkening the nymphs may be in no way related to the induction of the gregarious black patterning. Tanaka and Pener (1994) did not estimate the behavioural phase of the treated nymphs undergoing pigmentary changes and were unable to comment on the degree to which the phase state of the insect had been altered.

A recent report from Saini et al. (1995) indicates that a chemical signal, originating from the foam of egg pods, attracts gravid female S. gregaria to common egg-laying sites. Volatiles of egg pod foam were analysed by gas chromatography–antennography and gas chromatography– mass spectrometry, and two behaviourally active components were identified (Rai et al. 1997). Whilst these components may play a semiochemical role in attracting females to oviposition sites, there is no evidence to suggest that they alter the phase of the subsequent hatchlings in the manner suggested by the current findings. We believe the factor(s) affecting hatchling phase described here act in a wholly different manner and influence embryonic development.

Preliminary experiments to determine the nature of the gregarizing factor isolated in this study showed it to be a small (<3 kDa) hydrophilic substance, although solvents covering a full range of polarities were not used in these initial studies (Fig. 9, treatment 16; Fig. 13, treatments 23–26). The mode of action of the factor is at present unknown. Whatever the route of uptake by the egg, the factor clearly modifies embryonic gene expression and thus provides the hatchling with the most appropriate phase characteristics. In addition to preventing desiccation of eggs and facilitating the scramble of hatchlings to the surface, the chemical present in the foam plug could diffuse around and into the eggs of the pod during the initial part of the incubation period, thus triggering the switch for gregarization. Whilst the foam from egg pods laid by crowd-reared females could induce a large degree of gregarization in egg pods from solitary-reared females (Figs 7, 8, treatment 12), the reverse (i.e. solitarization) was not induced in the hatchlings when eggs from pods laid by gregarious crowd-reared females were treated with foam plugs from solitary females (Fig. 8, treatment 13). The foam of gregarious egg pods must therefore contain a factor which promotes or maintains gregarization in hatchlings from, for example, pods laid gregariously. Such a system of chemically signalling the phase of the next generation combines a positive cue for gregarization with a passive return to solitarization depending on maternal oviposition conditions.

The ability of the female locust to assess the probability that her offspring will emerge into a high-density population and direct phase development accordingly is remarkable. The definitive chemical identification of the gregarizing factor(s), its source within the reproductive tract and the control of its production are now under way in our laboratories and should provide clear details of the mechanism by which the female locust predisposes the phase characteristics of the hatchling. The identification of such a material has very considerable implications for locust control, and there is every possibility that agonists or antagonists of the factor or its production might provide the basis of new strategies to control locusts.

We thank Samantha James, Steve Roberts and Helen Oakley for maintaining the locust cultures. M.S.I. is grateful to the University of Rajshahi, Bangladesh, for granting him study leave. This work was funded by grant C-91346 from UNDP.