Table Of ContentsNext Page

Forecasting attacks by pest insects of cruciferous crops

Rosemary H Collier and Stan Finch

Horticulture Research International, Wellesbourne, Warwick CV35 9EF UK
Corresponding author: rosemary.collier@hri.ac.uk

Abstract

The timing of pest insect attacks can vary greatly from region to region and from year to year. A simulation method, based on rates of insect development, has been developed for forecasting the timing of insect attacks on cruciferous crops. The method is based on using a fixed number of individuals from one generation to the next and simulates the timing of events in the life cycle of the pests rather than their population dynamics. Forecasts produced for the cabbage root fly, the bronzed-blossom beetle and various species of Lepidoptera have been validated using pest monitoring data. Forecasts can be generated on either a regional basis from standard meteorological data, or on a local basis from air and soil temperatures collected by participating growers.

Introduction

Pest insects of horticultural crops are often controlled by spraying insecticide onto established crops. Such sprays are used against root-feeding insects because the insecticides applied at drilling or planting have usually degraded by the time the later generations of the pest become active. Sprays are also the simplest way of controlling foliar pests such as aphids and caterpillars. As the majority of insecticides recommended currently are of relatively short persistence, treatments are most effective if they are targeted to coincide with periods of peak pest activity. Unfortunately, the timing of such peaks can vary considerably from region to region and from year to year. Although it is possible to monitor the activity of many pest species using insect traps, routine monitoring is laborious and often requires specialist knowledge.

An alternative is to use weather data to forecast the timing of pest attacks. Forecasting systems have been developed for many insects. Many of these forecasts have been based on accumulated day-degrees (e.g. Eckenrode & Chapman 1972, Butts & McEwen 1981). However, day-degree forecasts have severe limitations, as their accuracy is based on the assumption that the relationship between the rate of insect development and temperature is strictly linear (Baker 1980). In addition, day-degree forecasts can be used only to predict the start and/or the peak of activity of the population. They cannot readily predict the spread of activity nor can they cope easily with insect populations that have polymodal patterns of activity. For example, the cabbage root fly (Delia radicum) can occur as one of two developmental biotypes, that emerge either `early' (April-May) or `late' (June-July) in the season (Finch & Collier 1983; Finch et al. 1988). As a result, the population of cabbage root flies within a particular locality, may consist primarily of one biotype or be a mixture of the two. Further problems in using day-degree models occur when attempts are made to include periods of diapause or aestivation, the major phases of insect dormancy that are induced by changes in temperature or photoperiod. As it is usual for only a proportion of the insect population to respond at any one time to a particular environmental cue, there is always variation between individuals in their rates of development. This is true of cabbage root fly populations during both aestivation and diapause (Collier & Finch 1983, Finch & Collier 1985).

At Horticulture Research International, a simulation method, based on rates of insect development and which allows for variation within the individuals in the insect population, has been produced for forecasting the timing of attack by a number of pest insects (Phelps et al. 1993). The simulation method has been used to develop forecasts for the cabbage root fly, bronzed-blossom beetles (Meligethes spp.) and more recently for some of the pest Lepidoptera of Brassica crops. The cabbage root fly and bronzed-blossom beetle forecasts are now used by growers. Similar forecasts for caterpillars are being validated currently. The biological basis, validation and practical uses of such forecasts are discussed in this paper.

The model

The forecasts were developed using a Monte Carlo simulation method (Phelps et al. 1993). The method uses a fixed number of individuals (usually 500, to obtain repeatable simulations) from one generation to the next and simulates the timing of events rather than the population dynamics of the insects. To develop each model, individuals at each stage of development (egg, larva, pupa, adult) were reared in cooling incubators at a range of constant temperatures between 6 and 30oC. The data recorded were used to determine the relationship between the rate of insect development and temperature. Linear or non-linear (Gompertz) curves were fitted to these data to provide equations, which could be incorporated into the model. In addition, variability was incorporated using the 'same-shape property' (Sharpe et al. 1977, Shaffer 1983). This implies that the coefficient of variation of the rate of insect development is constant at all temperatures. Account was taken also of periods of dormancy (aestivation and diapause) and of activity thresholds which might affect the outcome of the forecasts.

Ideally, the forecasts should be run using daily maximum and minimum air temperatures and maximum and minimum soil temperatures at a depth of 6-10 cm. However, maximum and minimum soil temperatures are not available from standard agro-meteorological stations in the UK. Therefore, the program uses several equations to estimate soil maximum and minimum temperatures from the air maximum, air minimum and 10 cm soil temperatures recorded daily at 09.00 h GMT (Phelps et al. 1993). If data for soil maximum and minimum temperatures are available, then the forecasts could be run equally well using these.

In general, the forecasts are run using accumulated temperatures for which the accumulation is started usually on 1 February each year. This is because late January is the natural break in insect development as it is usually the period of the year when temperatures are at their lowest. When temperatures start to rise again in the spring most of the species that overwinter in diapause have completed their diapause development and are ready to start post-diapause development. The diamondback moth (Plutella xylostella) is an exception. This species does not appear to overwinter successfully in large numbers in the UK (R. Collier, unpublished data) and the majority of insects are migrants from Europe and North Africa. Thus the forecast model has to be triggered by the arrival of migrant moths.

Forecast validation

The forecasts are validated using insect monitoring data. Cabbage root flies are monitored using either yellow water traps (Finch & Skinner 1974) or by sampling for fly eggs (Finch et al. 1975); bronzed-blossom beetles by using yellow sticky traps (Finch et al. 1990) and pest Lepidoptera by using yellow water traps, pheromone traps (for moth species) and by sampling plants for eggs and larvae.

Cabbage root fly

The cabbage root fly forecast (Collier et al. 1991) was developed originally for timing the application of mid-season insecticide treatments to control the fly on long-season Brassica crops such as swedes. Other uses of the forecast include warnings of the likely onset of third generation attack to Brussels sprout buttons and to autumn-sown crops of oilseed rape. At present, most insecticide treatments to leafy brassicas are applied prophylactically, before or soon after transplanting, and treatment against subsequent generations of this fly is usually unnecessary. However, the forecast could be used to indicate 'windows' where treatments would not be required. Figure 1 shows a comparison of the observed and forecast cabbage root fly activity at Kirton, Lincolnshire in 1999.

Figure 1. Comparison of the numbers of cabbage root fly eggs laid at Kirton, Lincolnshire in 1999 with a forecast of egg hatch generated using data from the nearby weather station.

Local variations in cabbage root fly activity include the co-existence in certain regions of the two developmental biotypes, with diapause of different durations (Finch & Collier 1983; Collier et al. 1989). Late-emerging flies emerge several weeks later than early-emerging flies so that, in effect, the generations of the two biotypes alternate. Similar damage to brassicas is caused by the closely-related turnip fly (Delia floralis) in Scotland and in some areas of south-west Lancashire (Finch et al. 1986). The presence of the two cabbage root fly biotypes and turnip fly in areas of south-west Lancashire means that in that specific locality there is continuous root fly pressure to Brassica crops throughout the summer. The cabbage root fly model produces forecasts for populations that contain specified proportions of the two biotypes. A turnip fly forecast has not yet been developed.

The cabbage root fly model has also been used to predict the changes that might occur as a result of global warming (Collier et al. 1990) and to predict cabbage root fly phenology in Spain, an EU member country in which calabrese production for the UK market can be affected severely by damage done to the crop florets by fly larvae.

Bronzed-blossom beetle

Feeding by adult bronzed-blossom beetles in mid-summer damages the curds or florets of cauliflower and calabrese (Finch et al. 1990) so that spray treatments are sometimes necessary. The forecast is used to predict the emergence of adult beetles from pupae within the previous host crop; usually oilseed rape. However, beetle infestations are not inevitable and seem to depend both on the proximity of oilseed rape crops and on the occurrence of warm, humid conditions during the main period of beetle migration. Figure 2 shows comparisons of observed and forecast bronzed-blossom beetle activity at Kirton in Lincolnshire in 1996.

Figure 2. Comparison of the numbers of bronzed-blossom beetles (Meligethes spp.) captured on sticky traps at Kirton, Lincolnshire in 1996 with a forecast of beetle emergence generated using data from the nearby weather station.

Caterpillars

In the UK, the caterpillars of several species of butterfly and moth can damage cruciferous crops. However, attacks by caterpillars tend to be sporadic and so do not occur in every crop every year. Considerable savings can be made in applications of insecticides for caterpillar control by applying sprays only when there are sufficient insects in the crop to warrant treatment. Forecasts of the development of the diamondback moth, small white butterfly (Pieris rapae L.), cabbage moth (Mamestra brassicae L.) and garden pebble moth (Evergestis forficalis L.) have been developed. In the UK, the diamondback moth and small white butterfly are the most important and widespread pest species. Because the diamondback moth is a migrant, pheromone traps are the most effective way of determining when infestations will occur, and the current diamondback moth model uses moth counts to trigger a forecast of subsequent population development. The other species overwinter well in the UK and forecasts for the small white butterfly and garden pebble moth gave a good indication of the periods when caterpillars were likely to be found in Brassica crops. It was not easy to validate the cabbage moth forecast, because caterpillar numbers were so low. Figures 3-5 show comparisons of the observed and forecast activity of the diamondback moth, the small white butterfly and the garden pebble moth at sites in Lincolnshire.

Figure 3. Comparison of the numbers of diamondback moth caterpillars found on insecticide-free Brussels sprout plants at Holbeach, Lincolnshire with forecasts of egg hatch generated using data from the nearby weather station. The forecasts were triggered using pheromone trap captures during the ‘first’ generation.

Figure 4. Comparisons between observed and forecast small white butterfly generations (total for three sites in south Lincolnshire). The forecasts were generated using data from the nearby weather station at Kirton. The horizontal lines with arrows indicate the periods when caterpillars would be expected from the forecast.

Figure 5. Comparisons between observed and forecast garden pebble moth generations at Holbeach, Lincolnshire in 2000. The forecasts were generated using data from the nearby weather station. The horizontal lines with arrows indicate the periods when caterpillars would be expected from the forecast.

Forecast validation

The current forecasts have been validated against as many sets of insect monitoring data as possible. The timing of pest activity may vary by 3-5 weeks between years and the numbers of generations may also vary. For example, the cabbage root fly may have a 'partial' third generation in warm locations, whilst completing only two generations in cooler areas such as Scotland. Use of the models has indicated that monitoring data must consist of >100 insects per generation if estimates of the timing of pest attacks are to be accurate to within one week (Collier & Phelps 1994). As the forecasts provide an indication only of the timing of pest attack and not of its severity, forecast data are generally expressed as percentages. When the times to 10% and 50% activity have been predicted and compared with the monitoring data, the majority of the pest forecasts have been accurate to within one week.

To date, forecasts for commercial growers have been produced using a network of rather widely-dispersed weather stations. However, there are obviously local differences in climate and in the degree of shelter, which might affect the timing of pest activity in a particular field. With very mobile insects such as the cabbage root fly, there may be little point in recording temperatures in individual fields, as the infesting population will have experienced the climate of the previous weeks, or months, in a different, unknown location. An intensive study in the Vale of Evesham (Finch & Skinner, unpublished data) indicated that there was very little difference in the timing of cabbage root fly activity from crop to crop. Although timing of activity may vary little within a region, intensive sampling in south-west Lancashire showed that the relative proportions of the two cabbage root fly biotypes varied considerably over relatively short distances (Finch et al. 1986).

At present, pest forecasts are based on Meteorological Office data collected from the network of weather stations, some of which are not particularly close to areas of commercial vegetable production. The actual forecasts are projected forwards using weather data from a previous, warm year. Since 1991, forecasts of cabbage root fly and bronzed-blossom beetle activity, based on weather data obtained from approximately 40 weather stations throughout the UK, have been made available to growers and advisers. Forecasts have been sent to growers each week for several weeks before and during the period of pest activity. The forecast models are now also available as part of a software package (MORPH), so that forecasts can be generated using growers' own weather data.

Future developments

The use of less-persistent insecticides, together with pressure from consumers and retailers to reduce the number of insecticide treatments applied to crops, means that growers need to target insecticide treatments more accurately. The next logical step, after determining the timing of pest activity, is to develop treatment thresholds to determine which of the various treatments are actually necessary. However, a considerable amount of further basic research will be required if we are ever to produce robust systems that will allow final crop damage to be forecast accurately from the numbers of insects monitored during the early stages of crop infestation.

Acknowledgements

We thank the UK Department of the Environment, Food and Rural Affairs (DEFRA) and the Horticultural development Council for supporting this work. We thank also participating growers, Sally Minns, Marian Elliott and Wendy Riggall of HRI and Dr Jennie Blood Smyth of ADAS for their help in collecting the monitoring data; and the Meteorological Office for providing the weather data.

References

Baker CRB. 1980. Some problems in using meteorological data to forecast the timing of insect life cycles. EPPO Bulletin 10, 83-91.

Butts RA & McEwen FL. 1981. Seasonal populations of the diamondback moth, Plutella xylostella (Lepidoptera: Plutellidae), in relation to day-degree accumulation. Canadian Entomologist 113, 127-131.

Collier RH & Finch S. 1983. Effects of intensity and duration of low temperatures in regulating diapause development of the cabbage root fly (Delia radicum). Entomologia experimentalis et applicata 34, 193-200.

Collier RH, Finch S & Anderson M. 1989. Laboratory studies on late-emergence in the cabbage root fly (Delia radicum). Entomologia experimentalis et applicata 50, 233-240.

Collier RH, Finch S, Phelps K & Thompson AR. 1990. Possible impact of global warming on cabbage root fly (Delia radicum) activity in the UK. Annals of Applied Biology 118, 261-271.

Collier RH, Finch S & Phelps K. 1991. A simulation model for forecasting the timing of attacks of Delia radicum on cruciferous crops. EPPO Bulletin 21, 419-424.

Collier RH & Phelps K. 1994. Carrot fly monitoring as an effective tool for pest management: how many flies have to be trapped? Aspects of Applied Biology 37, 259-263.

Eckenrode CK & Chapman RK. 1972. Seasonal adult cabbage maggot populations in the field in relation to thermal unit accumulations. Annals of the Entomological Society of America 65, 151-156.

Finch S & Skinner G. 1974). Some factors affecting the efficiency of water traps for capturing cabbage root flies. Annals of Applied Biology 77, 213-226.

Finch S, Skinner G & Freeman GH. 1975. The distribution and analysis of cabbage root fly egg populations. Annals of Applied Biology 79, 1-18.

Finch S & Collier RH. 1983. Emergence of flies from populations of overwintering cabbage root fly pupae. Ecological Entomology 8, 29-36.

Finch S & Collier RH. 1985. Laboratory studies on aestivation in the cabbage root fly (Delia radicum). Entomologia experimentalis et applicata 38, 137-14.

Finch S, Collier RH & Skinner G. 1986. Local population differences in emergence of cabbage root flies from south-west Lancashire: implicatons for pest forecasting and population divergence. Ecological Entomology 11, 139-145.

Finch S, Bromand B, Brunel E, Bues M, Collier RH, Dunne R, Foster G, Freuler J, Hommes M, Van Keymeulen M, Mowat DJ, Pelerents C, Skinner G, Stadler E & Theunissen J. 1988. Emergence of cabbage root flies from puparia collected throughout northern Europe. In: Progress on Pest Management in Field Vegetables (eds R Cavalloro & C Pelerents), P.P. Rotondo - D.G. XIII - Luxembourg No. EUR 10514. Balkema, Rotterdam, pp. 33-36.

Finch S, Collier RH & Elliott MS. 1990. Seasonal variations in the timing of attacks of bronzed-blossom beetles (Meligethes aeneus/Meligethes viridescens) on horticultural brassicas. In: Proceedings 1990 Brighton Crop Protection Conference - Pests and Diseases, 349-354.

Phelps K, Collier RH, Reader RJ & Finch S. 1993. Monte Carlo simulation method for forecasting the timing of pest insect attacks. Crop Protection 12, 335-342.

Shaffer PL. 1983. Prediction of variation in development period of insects and mites reared at constant temperatures. Environmental Entomology 12, 1012-1019.

Sharpe PJH, Curry GL, DeMichele DW & Cole CL. 1977. Distribution model of organisms development times. Journal of Theoretical Biology 66, 21-38.

Top Of PageNext Page