Introduced into Small Cabbage Plots

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Feb. 1983, p. 493-51 99-224/83/2493-9$2./ Copyright C 1983, American Society for Microbiology Vol. 45, No. 2 Dynamics of Baculovirus Growth and Dispersal in Mamestra brassicae L. (Lepidoptera:Noctuidae) Larval Populations Introduced into Small Cabbage Plots HUGH F. EVANS* AND GRAHAM P. ALLAWAYt Natural Environment Research Council, Institute of Virology, Oxford OX) 3SR, United Kingdom Received 25 June 1982/Accepted 18 October 1982 The nuclear polyhedrosis virus of Mamestra brassicae has been studied in larval populations of the moth introduced into small plots of cabbages. Primary dispersal of virus from single foci of infected larvae resulted from enhanced movement of the larvae, which colonized new plants logarithmically. Virus growth within the host population was quantified, and infection of young larvae in the following generation was related directly to the concentration of virus produced during the primary phase. Secondary cycling of virus resulted in dispersal of inoculum from multiple foci, and a large proportion of plants were ultimately colonized by infected larvae. The dynamics of virus growth during secondary dispersal were quantified and contrasted with results from the primary phase. The significance of these results is discussed in relation to possible control of insect pests through dispersal of virus by the host insect. Baculovirus diseases of insects are characterized by rapid replication (1), by the ability to spread within host populations, and by remarkable powers of persistence in the environment (9, 21). In general, persistence may be through the host insect (e.g., nuclear polyhedrosis virus [NPV] of Heliothis zea [12]), on plant surfaces (e.g., NPV of Orgyia leucostigma on balsam fir [3]), in the soil (e.g., NPV of Trichoplusia ni [14]), or through the medium of secondary hosts (passage of Lymantria dispar NPV by small mammals and birds [16]). Information on quantitative aspects of virus growth and disease spread and their relationships with host population density and distribution is limited. Natural epizootics of baculoviruses in forest pests have been described by Doane (2) (NPV of L. dispar) and Dahlsten and Thomas (1) (NPV of. pseudotsugata). Entwistle et al. (4-6) have described the epizootiology of an NPV disease in spruce sawfly (Gilpinia hercyniae) populations in which discrete host generations and a well-defined pattern of virus dispersal and persistence were components of the epizootic. Indeed, most epizootiological studies have been on forest pests, and fewer quantitative data are available on virus persistence and spread in which the host feeds on agricultural crops which t Present address: Memorial University of Newfoundland, St. John's, Newfoundland, Canada AlB 3V6. are periodically harvested, reducing the potential for persistence and dispersal. Payne et al. (2) and Tatchell (22) have described a series of experiments concerned with the granulosis virus of Pieris brassicae and have measured many of the dynamic variables of a system in which virus was applied as a spray to the crop plant to control the host insect. Data included larval feeding rates, duration of virus infectivity on the plants, and the dosage-mortality responses of the host to the virus, but no account was taken of virus dispersal or secondary cycling of virus within the host population. Spread of virus has received little attention. Ignoffo (13) advocated the release of infected larvae to spread virus through host populations, a method he termed "autodissemination." This term was also used to describe dispersal of NPV through contamination of adult Colias eurytheme genitalia (18). The aim of the present study was to quantify the role of NPV in the population dynamics of discrete populations of Mamestra brassicae larvae in small plots of cabbages. Laboratory studies had demonstrated that infected M. brassicae larvae showed an enhanced level of activity compared with healthy larvae, and this was a contributory factor in movement of larvae between cabbage plants. Detailed information on dosage-mortality responses (7) and virus productivity (9) in M. brassicae provided a quantitative background to the observations of NPV infection and distribution in field populations. 493

494 EVANS AND ALLAWAY APPL. ENVIRON. MICROBIOL. TABLE 1. Summary of design and sequence of experiments on dynamics of NPV in M. brassicae populations First phase Second phase Plot (yr) Layout (dispersal of larvae from discrete foci) (effects of residual inoculum from first phase) 1 (1979) 9 x 5 Virus-infected, third-instar larvae onto 5 Residual inoculum inducing primary incentral plants; daily observation of dis- fection; 5 neonate larvae onto each tribution until larval death plant; remove after 1 days and feed individually until virus death or larval pupation 2 (1979) 9 x 5 Same as plot 1 Nil 3 (198) 11 x 1 Virus-infected, third-instar larvae onto 6 Full expression of residual inoculum on central plants; observation of distribu- new larval generation with secondary tion every 2 days until larval death dispersal of infection; observe at intervals and score distribution, age, and level of infection of larvae 4 (198) 1 x 9 Healthy third-instar larvae onto 6 central Nil plants; observation of distribution every 2 days until pupation MATERIALS AND METHODS Layout of cabbage plots. Cabbages (variety, Sutton's May Express) at the 8 to 1-leaf stage were planted on regular grid spacing with.5 m between plants. Experiments on virus dispersal and primary virus utilization were carried out, in 1979, on 9 x 5 grid layouts (two plots). Primary dispersal and secondary cycling of virus were investigated, in 198, on an 11 x 1 grid layout, whereas a control plot with healthy larvae only was set up on a 9 x 1 grid. Cabbages were planted 7 days before the start of each experiment, and 2-cm mesh netting was used to exclude birds. Experimental procedure. A summary of the designs and sequence of experiments is shown in Table 1. Fuller details of the sampling regimes and rationale behind the experiments are given below. (i) First phase: primary dispersal from discrete foci of disease. Third-instar M. brassicae larvae were fed in the laboratory with lethal dosages of NPV. On day 3 after infection, 15 larvae were placed on each of five (45-plant experiment) or 6 (11-plant experiment) central plants within each virus plot (Fig. 1, plots 1, 2, and 3). Fifteen healthy third-instar larvae were placed on six central plants in the control plot (Fig. 1, plot 4). Observations were carried out by inspecting all leaves on each plant at 1-day (45-plant plots) or 2-day (11- and 9-plant plots) intervals, with care being taken to minimize disturbance of feeding larvae. Observations included the number, age, and position of larvae (both on and between plants) and visual diagnosis for presence of NPV (healthy, infected, or virus killed, using easily recognizable symptoms of disease previously defined from laboratory studies). Observations were carried out until all larvae had died or pupated. (ii) Second phase: effects of residual inoculum from first phase. The fate of NPV polyhedra produced during the primary dispersal phase was investigated in two ways. (a) Effectiveness as residual inoculum inducing primary infection in a succeeding host generation. The aim of this experiment was to quantify early mortality resulting directly from inoculum remaining from the previous host generation. This experiment was carried out in 45-plant plot 1. Fifty newly hatched first-instar larvae were placed on each plant and left for 1 days, which at field temperatures was known to be a sufficient time for larvae to become infected but not long enough for deaths and secondary cycling of virus to occur. All larvae were removed by destructive sampling of the plants and were placed in individual closed pots containing semisynthetic diet. Larvae were inspected daily until all had died or pupated, when all were diagnosed for the presence of NPV by Giemsastained smears examined at x1,8 magnification. Effectiveness as primary inoculum was expressed as the proportion of larvae found on each plant that died from NPV disease. (b) Effectiveness of residual inoculum on a succeeding host generation where secondary cycling of virus takes place. The experimental procedure was similar to the early stages of (a) above, but larvae were left on the plants throughout the experiment. This was designed to measure the impact of both primary and any secondary inoculum produced on an initially healthy host population. Each plant in the 11-plant plot was infested with 25 neonate larvae, and sampling was carried out by observation of all plants for number, position, age of larvae, and presence of NPV. Observations were carried out at approximately 4-day intervals, but poor weather resulted in extended sampling intervals during the later stages. The experiment was terminated when all larvae had died or pupated. Subsidiary field observations. Pitfall traps (wet type) for ground-dwelling predators, particularly carabid beetles, were placed at 2-m intervals within and up to 6 m outside the plots. Predators and M. brassicae larvae were removed on each sampling occasion and inspected for NPV polyhedra, using Giemsa-stained smears of midguts and whole larvae, respectively. Laboratory experiments to determine absolute rates of larval movement away from cabbage plants. Different densities of M. brassicae eggs were placed on individual cabbage plants in a constant-environment chamber programmed to mimic the temperature (12 to

VOL. 45, 1983 BACULOVIRUS GROWTH IN M. BRASSICAE L. 495 plot 4 ;control o TlI I I A L1 -L I o * @ A EC U A 1 1 A A U U A A I T 1o 1 11 plot I 3 ;virus ~~~ A A A a A a A * A * A U @ A A * A plot 2 ;virus J * m A v = * plot 1 _ A A._A _1_I. O j j A A A U A1 ;virus AX*TI*T*T U I1 11 1- loi * A FIG. 1. Plan views of the primary dispersal phase in virus plots 1 to 3 and control plot 4. The symbols represent the distribution of M. brassicae larvae at different times after plant infestation (number of days): plot 1-(4) 1, (-) 2, (A) 3, () 4, () 5 to 8; plot 2-4() 1, (M) 2, (A) 3, () 4, () 5, (A) 6, (V) 7; plot 3-4) 1, (A) 3, (A) 5 to 9; plot 4-(A) 1, (M) 3, (A) 5, () 7, () 9. 24 C), humidity (5 to 9%o relative humidity), and light patterns (18-h day) observed in the field at the time. Half the plants were surface contaminated with virus by smearing both surfaces of the leaves with NPV suspension; the other half served as virus-free control plants. Each plant was placed in water (2-cm depth) containing a few drops of detergent. Larvae hatching from the eggs and leaving the plants were trapped in the water, thus providing a measure of absolute rates of movement away from the plants. Water troughs were inspected daily, and larvae found were screened for the presence of NPV. This provided comparison of larval movement rates at different densities and between healthy larvae and those becoming infected after ingestion of virus. RESULTS Primary phase. (i) Larval dispersal from discrete foci. The rate and extent of larval movement away from the central plants were obtained from the observations of larval distribution. Plan views of the four experimental plots are shown in Fig. 1, with the different symbols indicating the time course of observations. In all cases there was a rapid expansion of the overall distribution of larvae with some evidence of directional movement in the NE/SW plane. Stability of larval distribution from day 5 in the virus plots reflected the reduced activity before death from NPV. At this stage larvae exhibited typical symptoms of NPV infection (white shiny appearance after movement from the hearts of the cabbages to the outer leaves), and all eventually died on the exposed upper leaf surfaces. In contrast, healthy larvae in the control plots continued feeding in the cabbage hearts and caused considerable damage before moving to new plants. In assessing larval transmission of virus between plants, it is appropriate to look at the accumulated totals of plants with larvae present since this reflects not only the larval totals but also their distribution. Figure 2 illustrates the way in which plants became colonized as the experiment progressed. The relationships were all of the same general logarithmic form (y = a + b logex). There was no significant difference between the respective slopes, indicating that absolute rates of dispersal were similar for different plots and for healthy and virus-infected larvae. However, the numbers of larvae initially establishing on the plants after early unexplained losses were markedly different between plots, which had a strong influence on the rates of dispersal. When these initial larval numbers were plotted against dispersal rates (the slopes,

496 EVANS AND ALLAWAY 4' 2_- 2 4 6 8 1 12 FIG. 2. Primary phase: relationships between accumulated numbers of cabbage plants colonized by M. brassicae larvae and days post-plant infestation. Relationships for each plot were as follows. Virus-infected larvae: (plot 1) y = 6.5 + 14.4 log,x (), r2 =.97; (plot 2) y = 2.72 + 13.69 log,x (), r2 =.97; (plot 3) y = 5.99 + 15.52 logex (A), r2 =.99. Healthy larvae: (plot 4) y = 6.72 + 13.64 log,x (A), r2 =.99. b, of the dispersal relationships), it was clear that there was a linear correlation between the two variables in the virus plots (Fig. 3, y = 11.3 +.75x; r2 =.98). The dispersal rate for the control plot was significantly lower (at 95% probability level, virus lower limit = 13.94; control value = 13.64) for the observed initial larval number (Fig. 3) and indicated that infected larvae were more likely to disperse than were healthy larvae for a given initial larval density. A similar relationship was observed in the laboratory experiments on rates of movement away from individual cabbage plants. Dispersal rates were obtained from the accumulated numbers of larvae migrating from the cabbage plants 16 3 d'* c N o 2 E 1 oo * 1. ~ ~ ~ ~ ~ ~ ~ 1.. 14/ -13. 2 3 4 5 6 initial larval no. FIG. 3. Primary phase: relationship between infected M. brassicae larval dispersal rates () (slope b of the relationship y = a + b logx [Fig. 2]) and total numbers of larvae initially present on the cabbage plants (y = 11.3 +.75x; r2 =.98). The position of the dispersal rate for healthy larvae is indicated (). 14 APPL. ENVIRON. MICROBIOL. " 4 8-12 initial larval no. FIG. 4. Relationships between larval dispersal rates and initial larval establishment for M. brassicae larvae on cabbage plants in the laboratory. Symbols: virus-infected larvae-4) y = -2.16 +.55x (r2 =.98); healthy larvae-(o) y = 1.66 +.24x (r2 =.97). F test (slope) = 35.29 significant at 99%o probability level. over the time course of the experiment. All larvae which left the virus-contaminated plants after day 2 were found to be infected with NPV. When the dispersal rates (slope values) were plotted against initial larval establishment, there were significant slope differences at the 99% level (F test) between the linear relationships for virus-infected and healthy larvae (Fig. 4). The dispersal rate for virus-infected larvae was 2.3 times higher than for healthy larvae. (ii) Dynamics of virus spread and release. Larval distributions within the plots gave a measure of total dispersal and hence allowed comparison of intrinsic dispersal rates between healthy and virus-infected larvae. In the context of effective dispersal of the virus itself, it is more important to assess the distribution of larvae dying from, and releasing, virus. Figure 5 shows the accumulated numbers of plants having virus-killed larvae during the primary phase. The general form of the relationship (y = a + bx2) differs from the dispersal rates for absolute plant colonization shown in Fig. 2. These data reflect the fact that larvae recorded during one sample are accounted for in the next sample and their distribution is therefore static. It is interesting to note that there was a lower rate of virus dispersal in plot 2 than in the other two virus plots. This probably reflected the lower initial larval establishment in this plot, limiting the number of plants which could be colonized. Virus growth in relation to larval age and time postinfection has been quantified for M. brassicae (1). These data were used to predict, from known larval densities and daily instar distributions, how many polyhedra were present in the cabbage plots during the experiment. Accumulated virus production (log1 polyhedra) has

VOL. 45, 1983 BACULOVIRUS GROWTH IN M. BRASSICAE L. 497 a 2 d a N>15 la,x,x 1 _ a = E X n.c 5._. 1 " ""', I--,' X/ 7- E a 4- a.3 +3-2 4 6 8 1 FIG. 5. Primary phase: relationships between accumulated numbers of plants with virus-killed M. brassicae larvae and days post-plant infestation. Symbols: plots 1 and 3 combined (v), y = -3.17 +.24x2 (r2 =.98); plot 2 (), y = -2.16 +.15x2 (r2 =.98). been plotted against time post-plant infestation in Fig. 6. Data from all virus plots have been combined since there was no significant difference between the individual plots, despite the initial differences in larval establishment. The logarithmic buildup of total inoculum within the plots was described by y = 5.8 + 2.2 logx (r2 -.94). (iii) Effectiveness of released virus inoculum. The data expressed in Fig. 6 gave the total observed virus productivity in the cabbage plots and, in epizootiological terms, provided a minimum estimate for the density of infective virus. More detailed information on the primary role of virus released from dead larvae was obtained in plot 2 by the bioassay described in method a (see above). It was assumed that any viral mortality observed in these neonate larvae was the result of feeding on inoculum persisting from the primary dispersal phase. Dosages on individual plants (based on observation of larval age at d c oc E. a J FIG. 6. Primary phase: relationship between accumulated total numbers of M. brassicae NPV polyhedra in the plots and days post-plant infestation. Plots 1 to 3 combined: log1ly = 5.8 + 2.2 logex; r2 =.94. i 8 F 1 dosage: log,o polyhedra per plant FIG. 7. Secondary phase: weighted linear regression of probit mortality of neonate M. brassicae larvae on log1o dosage of polyhedra remaining on the plants from the primary experiment. Probit y = -12.69 + 2.6 log1o polyhedra (x). Log1 5%'o lethal dose (wholeplant dosage) = 8.59; logl 95% limits = 8.67, 8.5. The 95% fiducial limits are shown. death from virus) and first-instar mortality from virus on these plants were used to calculate a weighted linear regression of probit mortality on loglo virus dosage (11). This is shown in Fig. 7 and illustrates that, although the heterogeneity of response was greater and despite the relatively crude bioassay, the slope of the relationship was close to that obtained in more controlled laboratory studies (7). Larvae on some plants which appeared to have no larval corpses from the primary phase were nevertheless subject to some virus mortality in the first-instar bioassay. Back-extrapolation from these mortalities, using the rearranged probit relationship [loglox = (probit y + 12.69)/2.6], indicates the theoretical dosages present on the plants which would have given rise to the observed mortalities. The data on whole-plant dosages were compared with the virus productivity of those larvae dying in the bioassay to calculate productivity ratios for primary inoculum. This statistic measures the efficiency of virus replication in relation to primary virus inoculum and, since higher larval densities inevitably result in greater virus productivity, in relation to initial larval establishment. Thus, when productivity ratios were plotted against larval establishment per plant, there was a strong linear correlation between the two variables (Fig. 8). Clearly, virus utilization as primary inoculum was inefficient below a calculated larval establishment of 37 per plant. Net productivity below this larval density was less than 1 and, without further means of virus replication, would eventually result in virus extinction. However, as will be shown later, this is

498 EVANS AND ALLAWAY APPL. ENVIRON. MICROBIOL. 1*2. X 8._, -4- la Q - - * S " 1 2 initial larval no. per 3 plant - 4 FIG. 8. Secondary phase: relationship between productivity ratio (= virus produced on larval death/ virus originally present on the plant) and initial larval establishment on the plant. y = -.8 =.3x; r2 =.88. 8 4.8 6 1. o 4 E. X2 co E U2. m - 2 4 6 8 FIG. 1. Secondary phase: relationship between accumulated numbers of plants with virus-killed M. brassicae larvae and days since larvae were placed on the plants. y = -.899 +.41 logex; r2 =.96. unlikely to happen since secondary infection of older larvae with the resultant greater productivity more than compensates for low productivity ratios in the first instar. Secondary phase. (i) Dynamics of virus spread and release. Virus utilization and the full expression of secondarily produced virus was studied in plot 3. The experiment enabled observations to be made on the distribution, age structure, and incidence of infection in a complete larval generation of the host. (i) Dispersal of larvae from multiple foci. As larvae became infected by ingestion of virus remaining from the primary phase, they dispersed from the many centers of disease present. As these larvae died, the resulting secondary inoculum acted as further sources of virus and infected larvae spread even further. The distribution of virus-infected larvae within the plot 3 ;second phase oi I A U U @ A A U A A A A A& A U A A A A U A U A A A A A S U A7 1^1A A I A A A O A :lol I lolo A A A A A A * A oo FIG. 9. Plan view of secondary dispersal phase. Distribution of virus-killed M. brassicae larvae in plot 3 against number of days since larvae were placed on the plants. Symbols represent the time course in days: () 9, (U) 12, (A) 16, () 19 to 26, (O) 41, (A) 46 to 6. plot is illustrated, in plan view, in Fig. 9, where the different symbols indicate the time course. Diseased larvae eventually colonized 78% of plants within the plot. (ii) Dynamics of secondary virus growth. As in the primary phase, a good measure of virus spread was the accumulative buildup within the plot of plants having virus-killed larvae. Figure 1 shows that, in contrast to the primary phase (Fig. 5), the rate of dispersal was logarithmic and presumably reflected the movement of infected larvae from a multiplicity of disease foci as well as the higher, evenly distributed population of susceptible larvae present. Accumulated total virus productivity within the plot was calculated from the distribution and age structure of virus-killed larvae. Total numbers of polyhedra increased logarithmically with time (Fig. 11) in a similar way to total plants colonized (Fig. 1). However, although this relationship was a reasonable predictor of the initial and final stages of virus accumulation, it did not account for the sharp rise between days 41 and 46. This was probably a result of the response to secondary rather than primary inoculum, since most of the larvae dying at this stage were in the sixth instar and must have ingested dosages well in excess of those available as primary inoculum. Growth characteristics of both virus accumulation and dead larval totals were similar, indicating a close relationship between the two. This is confirmed in Fig. 12 (top), where accumulated virus production (log1o polyhedra) increases linearly with accumulated plants having dead larvae. This is not surprising in view of the fact that the measure of inoculum buildup was derived from known productivities of virus-killed larvae. However, there was also a time element involved in this relationship since plant coloniza-

VOL. 45, 1983 d 12 1 83 2 6 E o la Or5 Cf t4 ~~~~~* ~ 2 4 6 8 FIG. 11. Secondary phase: relationship between accumulated total numbers of M. brassicae polyhedra in plot 3 and days since larvae were placed on the plants. log1ly = 4.45 + 1.73 logex; r2 =.95. BACULOVIRUS GROWTH IN M. BRASSICAE L. 499 b. d C X C 5* c - 31 E QL.1 Io 2 4 6 8 FIG. 13. Secondary phase: mean number of viruskilled larvae per plant at each sample after larval establishment. Each point represents larval mortality since the previous sample. tion proceeded sequentially, and therefore larvae were maturing as they dispersed. Thus, the amount of virus per plant was greater towards the end of the experiment, showing a proportionate daily rate of increase of.3 (Fig. 12, bottom). Obviously the quantity of virus per plant will reflect not only the absolute numbers of larvae present but also the virus productivities for those larvae. When the mean number of viruskilled larvae per plant was plotted against time, there was initially a remarkable constancy as larvae aged, became infected, and eventually died (Fig. 13). As the peak of the logarithmic rise in the number of larvae lethally infected was reached, there was a sudden rise in the number of dead larvae per plant (Fig. 13, day 41) which then stabilized at the higher level. This was probably a further reflection of larval response to large quantities of secondary inoculum. Ino. plants with dead larvae FIG. 12. Secondary phase: relationships between log1 accumulated M. brassicae NPV polyhedral production and accumulated numbers of plants with dead larvae. (top) Total polyhedra in plot 3-4) loglay = 8.27 +.4x (r2 =.99); (bottom) log1 polyhedra per plant-4o) logloy = 7.3 +.3x (r2 =.99). Pitfall traps. M. brassicae larvae collected from pitfall traps within the virus plots during the primary dispersal phase all had NPV disease. During the secondary dispersal phase few larvae with NPV disease were found, except towards the end of the experiment, when most dead larvae were present on the plants. Carabid beetles, notably Pterostichus madidus and Bembidion sp., were common in the plots and most traps yielded specimens. A few of these specimens were shown to have intact polyhedra in their guts. Bioassay of gut contents and feces against first-instar larvae indicated that these polyhedra were infective. This suggested strongly that carabid beetles had been feeding on infected M. brassicae larvae in the field. However, these data are incomplete and are being investigated further before firm conclusions are made. DISCUSSION The concept of autodissemination of virus by larval movement suggested by Ignoffo (13) applies clearly to these experiments. Both primary dispersal from discrete foci and secondary dispersal from multiple foci have been demonstrated in the small plots used. Indeed, the effect of secondary cycling of virus was to induce a dramatic increase in the overall distribution of NPV disease in the host population, resulting in 78% of plants carrying virus-killed larvae. Infected larvae moved at least 2.5 m from the center to the farthest edges of the 11-plant plot, and it is clear that this is an underestimate of their maximum dispersal range since some larvae were found in pitfall traps 2 m outside the experimental plot. The epizootiology of M. brassicae NPV in these plots was governed by the effectiveness of inoculum produced after primary dispersal of larvae. Large quantities of polyhedra were re-

5 EVANS AND ALLAWAY leased by disintegration of virus-killed larvae on the plants. However, the resultant mortality of neonate larvae placed on these plants, although related directly to the amount of this inoculum (Fig. 7), was much lower than expected from a laboratory study which showed the first-instar 5% lethal dose to be seven polyhedra (7). Calculated dosages on the plants represented enormous multiples of the first-instar 99% lethal dose, thus showing that in numerical terms virus was relatively poorly utilized. The apparent lack of effectiveness of the inoculum may be attributable to low encounter rates between neonate larvae and the virus sources. Virus was relatively localized on the plants, and its distribution tended to reflect the configuration of leaves. In this way virus released by disintegration of larvae, which typically died on the topmost edges of leaves, was washed vertically downwards, its final distribution being influenced by the topography of the leaves themselves. Young larvae fed on small areas on the undersides of outer leaves and had low feeding rates so that the possibility of encounter between host and virus depended on (i) the initial distribution of host and virus and (ii) whether host or virus distribution changed to bring both together. Biodegradation or physical loss of virus will reduce the amount of primary inoculum present as the host population matures. Up to 9% of M. brassicae polyhedra released on cabbage plants were shown to accumulate in the soil below (8), although a proportion of these may have been available for inducing infection since, as suggested by Jaques (15), soil may act as an effective reservoir for virus. Payne (19) showed that purified granulosis virus capsules on cabbage adhered strongly but lost their activity rapidly, possibly indicating that even when partially protected by larval remains virus released onto cabbage surfaces may retain activity for a short time only. During the secondary phase larval maturation had an effect on the potential role of virus present on the plants. First, maturation resulted in a large decrease in susceptibility (7), thus requiring larvae to ingest greater quantities of a decreasing source of virus whose distribution may have differed from that of the host. Second, larvae had an increasing food requirement in each instar, culminating in large-scale damage to leaves from feeding by fifth- and sixth-instar larvae. The greater feeding rate would, however, increase the probability of encountering localized sources of virus and may have compensated for the decreased susceptibility in older larvae. In this case the result was a large increase in numbers of larvae infected in the fifth and sixth instars as encounters between the host and both residual primary and secondary inocula APPL. ENVIRON. MICROBIOL. increased (Fig. 9, 1, 13). Changes of behavior during the course of NPV infection are common in lepidopterous larvae (17; H. F. Evans, unpublished data). Enhanced activity of infected larvae, particularly the tendency to migrate to the upper edges of leaves, was clearly a major factor in the spread of NPV disease within the cabbage plots in the experiments described here. Although not demonstrated conclusively, the presence of infective virus in the guts of predatory beetles within the plots indicated a further potential method for spreading virus and might provide a partial explanation for the mortality of neonate larvae on plants where no virus source was identified. Autodissemination clearly has some potential for initiating epizootics in field populations of crop pests. However, the data presented here suggest that it would not be practical where host populations are low, even though considerable virus spread occurs under these circumstances. Also, the maximum effect of dispersed virus was manifested in the fifth and sixth instars when cabbage plants had already been heavily damaged. More virus-infected larvae would have to be released, and the susceptible host population would have to be higher, for the encounter rate between host and virus to be sufficient to ensure successful control by autodissemination. These are precisely the conditions which Ignoffo (13) described in his experimental evaluation of autodissemination. Further research is necessary to test the feasibility of this method of pest control under less ideal conditions. These experiments form part of a series on the dynamics of a virus-host interaction involving, as models, M. brassicae and its NPV. The data presented here show that even under simple experimental conditions in the field the role of NPV in M. brassicae population dynamics is a complex one. ACKNOWLEDGMENTS We thank J. S. Robertson and P. F. Entwistle for constructive comments during the preparation of this manuscript. LITERATURE CITED 1. Dahlsten, D. L., and G. M. Thomas. 1969. A nucleopolyhedrosis virus in populations of the Douglas-fir tussock moth, Hemerocampa pseudotsugata, in California. J. Invertebr. Pathol. 13:264-271. 2. Doane, C. C. 1976. Ecology of pathogens of the gypsy moth, p. 285-293. In J. F. Anderson and H. K. Kaya (ed.), Perspectives in forest entomology. Academic Press, Inc., New York. 3. Elgee, E. 1975. Persistence of a virus of the white-marked tussock moth on balsam fir foliage. Bi Mon. Res. Notes Can. For. Serv. 31:33-34. 4. Entwistle, P. F., P. H. W. Adams, and H. F. Evans. 1977. Epizootiology of a nuclear-polyhedrosis virus in European spruce sawfly (Gilpinia hercyniae): the status of birds as dispersal agents of the virus during the larval season. J. Invertebr. Pathol. 29:354-36.

VOL. 45, 1983 BACULOVIRUS GROWTH IN M. BRASSICAE L. 51 5. Entwiste, P. F., P. H. W. Adams, and H. F. Evans. 1977. Epizootiology of a nuclear-polyhedrosis virus in European spruce sawfly Gilpinia hercyniae: birds as dispersal agents of the virus during the winter. J. Invertebr. Pathol. 3:15-19. 6. Entwistle, P. F., P. H. W. Adams, and H. F. Evans. 1978. Epizootiology of a nuclear polyhedrosis virus in European spruce sawfly (Gilpinia hercyniae): the rate of passage of infective virus through the gut of birds during cage tests. J. Invertebr. Pathol. 31:37-312. 7. Evans, H. F. 1981. Quantitative assessment of the relationships between dosage and response of the nuclear polyhedrosis virus of Mamestra brassicae. J. Invertebr. Pathol. 37:11-19. 8. Evans, H. F. 1982. The ecology of Mamestra brassicae NPV in soil. Proc. Int. Colloq. Invertebr. Pathol., 3rd, p. 37-312. 9. Evans, H. F., and K. A. Harrap. 1982. Persistence of insect viruses, p. 54-96. In B. W. J. Mahy, A. C. Minson, and G. K. Darby (ed.), Virus persistence. Symp. 33, Society for General Microbiology Ltd. Cambridge University Press, London. 1. Evans, H. F., C. J. Lomer, and D. C. Kelly. 1981. Growth of nuclear polyhedrosis virus in larvae of the cabbage moth, Mamestra brassicae L. Arch. Virol. 7:27-214. 11. Flnney, D. J. 1971. Probit analysis, 3rd ed. Cambridge University Press, London. 12. Hamm, J. J., and J. R. Young. 1974. Mode of transmission of nuclear-polyhedrosis virus to progeny of adult Heliothis zea. J. Invertebr. Pathol. 24:7-81. 13. Ignoffo, C. M. 1978. Strategies to increase the use of entomopathogens. J. Invertebr. Pathol. 31:1-3. 14. Jaques, R. P. 197. Natural occurrence of viruses of the cabbage looper in field plots. Can. Entomol. 12:36-41. 15. Jaques, R. P. 1975. Persistence, accummulation, and denaturation of nuclear polyhedrosis and granulosis viruses, p. 9-11. In M. Summers, R. Engler, L. A. Falcon, and P. Vail (ed.), Baculoviruses for insect pest control: safety considerations. American Society for Microbiology, Washington, D.C. 16. Laublatr, R. A., mad J. D. Podgwte. 1979. Passage of nucleopolyhedrosis virus by avian and mammalian predators of the gypsy moth, Lymantria dispar. Environ. Entomol. 8:21-214. 17. Lewis, F. B. 197. Mass propagation of insect viruses with specific reference to forest insects. Proc. Int. Colloq. Insect Pathol., 4th, College Park, Md., p. 32-326. 18. Martignoul, M. E., and J. E. Mdistead. 1962. Trans-ovum transmission of the nuclear polyhedrosis virus of Colias eurytheme Boisduval through contamination of the female genitalia. J. Insect Pathol. 4:113-121. 19. Payne, C. C. 1982. Insect viruses as control agents. Parasitology 84:35-77. 2. Payne, C. C., G. M. Tatcbell, and C. F. Williams. 1981. Comparative susceptibilities of Pieris brassicae and P. rapae to a granulosis virus from P. brassicae. J. Invertebr. Pathol. 38:273-28. 21. Tanada, Y. 1971. Persistence of entomogenous viruses in the insect ecosystem, p. 367-379. In Entomological essays to commemorate the retirement of Professor K. Yasumatsu. Hokuryukan, Tokyo. 22. Tatchell, G. M. 1981. The effects of a granulosis virus infection and temperature on the food consumption of Pieris rapae. Entomophaga 26:291-299.