Abstract

We modelled the population phenology of the diamondback moth (DBM), Plutella xylostella (L.) (Yponomeutidae) with a simple degree day model implemented using DYMEX, a population dynamics modelling program. Predicted and observed developmental periods of DBM reared under field conditions during the months of October 1999 to February 2000 at Forthside Research Station, Tasmania (which is in the temperate zone) were compared. Threshold temperature values of the second and third instar larvae were modified using values from previously published data to obtain a marginally closer fit to the observed developmental periods. The model overestimated the developmental periods when standard daily Stevenson’s screen temperatures were used. Better predictions were obtained using daily temperatures recorded at 20 cm above ground level. We test this further by comparing observed and predicted development for DBM cohorts reared on potted Chinese cabbage under open, semi-shade, fully shaded and glasshouse conditions during August to November 2000 in Brisbane (which is in the sub-tropics). Temperatures recorded on the plant surface gave good predictions of development period from egg to adult emergence for all cohorts. The development model needs to be tested under a wider range of conditions, but our results suggest DBM development can be readily modelled. This work can form the basis of attempts to forecast DBM population dynamics and interpret population phenology.

Keywords

population phenology modelling, DYMEX, developmental rate, degree days, micro-habitat temperatures

Introduction

The diamondback moth (DBM), Plutella xylostella (L.) (Lepidoptera: Yponomeutidae) is a serious pest of cruciferous crops around the world. A plethora of research has been initiated on the management of DBM. Various pest management strategies have been developed and central to these is an understanding of the species' population phenology and dynamics.

Fundamental to modelling a species' phenology is an understanding of the insects' development rate in relation to temperature. From such relations the species phenology in an area can be predicted. We use DYMEX, an interactive population dynamics modelling program which predicts the relative abundance of an organism over a given period and allows rapid evaluation of management options (Maywald et al. 1999), to model DBM development.

Stanaway and Houlding (2000) built a DYMEX model for DBM on Brassica crops under field conditions. The model employed temperature driven cohort development based on studies of Goodwin (1976) and development was described by a linear above threshold function. Model predictions were compared with DBM larval counts obtained from Gatton Research Station Queensland (27° 33' S, 152° 16' E) over a three year period (1992-1995) and its overall fit was reported as good, despite problems encountered with over estimations of relative abundance (Stanaway & Houlding 2000). To further validate the model we compared the development periods of DBM obtained from field rearing at Forthside Research Station in Tasmania (41° 21'S, 146° 26' E) and a range of conditions in Brisbane, Queensland (27°28' S, 153° 02' E), with the predictions of the model. We show that with appropriate development rate functions and associated temperatures DBM development can be readily modelled.

Methods

Experiment 1

Development data obtained from rearing DBM on thirteen potted broccoli plants at Forthside Research Station during the months of October 1999 to February 2000 were used to test the model (Hill et al. unpublished). Locally reared adults were allowed to lay eggs on leaves of potted broccoli plants (except where wild eggs collected from turnip cotyledons were placed on two plants) and development period from egg to adult emergence was recorded. In two cases developmental period from neonate larvae to adult emergence was measured. The experimental plants were caged around the onset of the pre-pupal stage and day of emergence of adult moths inside the cage recorded.

Minimum, maximum and average temperatures for the period of 1 October 1999 to 8 March 2000, recorded at Forthside Research Station were used to drive development in the model. Best fitting predictions were obtained in three steps. Firstly, predictions yielded by using unchanged, default development parameter values were compared with the observed development periods. In step two, the default parameter values affecting development in II and III instar larvae were modified based on previously published data to obtain a marginally close fit. Thirdly, using the changed development parameters, better predictions were obtained when temperatures recorded 20 cm above the ground level during the experimental period were used to drive the model.

Experiment 2

Cohorts of DBM were reared on potted Chinese cabbage under glasshouse, open, semi-shade and fully shaded conditions thrice from August to November 2000 in Brisbane. We used DBM adults from the Gatton Research Station culture caged over potted plants. Newly laid eggs were thinned to 8 eggs per plant. There were 6 plants per treatment. Development of immatures and temperatures were recorded twice daily; once early in the morning (0500-0700) and again after mid-day (1300-1500). Temperatures were recorded using an optical infra-red thermometer (RayTek-Pmplus ™, Santa Cruz, USA) at a number of locations where immature stages were found on each plant. By keeping plants under a range of cage conditions we generated a number of distinct temperature regimes at the one time. Time of adult emergence was recorded. We used a simple degree-day model based on thresholds for the entire immature period and compare this prediction with the observed number of degree-days required for development.

Results

Experiment 1

Predictions of developmental periods, yielded when default parameter values were used, over-estimated the observed developmental periods in the range of 4 to 87% (Table 1). After changing the threshold temperature values to 8°C from 13°C and 11.5°C in II and III instar larvae respectively and the temperature slope values to 0.05 from 0.065 in both the above instars, marginally close predictions were obtained (Prediction II, Table 1). The difference of this prediction with the observed was in the range of -13% to 71%. When simulations were run using adjusted parameter values and temperatures recorded 20 cm above ground level the predictions differed from the observed in the range of –27 to 26%, with a mean difference of only 3% (Prediction III, Table 1). In this case, four predictions of development periods matched exactly with the observed and three differed from the observed in the range of 4 to 8%.

Experiment 2

By placing cohorts in a range of microclimatic conditions, we varied both the temperatures experienced and the range of development times recorded at the one time of year (Figure 1, Table 2). Using a simple degree-day model for the entire immature period (egg to adult) with a developmental threshold of 7°C and the temperatures recorded for each cohort, we could readily predict the mean development period of DBM (Table 2).

Table 1. Comparison of mean development time of DBM in days observed on 13 plants at Forthside, Tasmania and that predicted using a simple degree day model implemented using DYMEX. Prediction I is based on development parameters taken from Goodwin (1976) and Stevenson screen temperatures. Prediction II is as in I, but has adjustments to some thresholds and rates (see text for details). Prediction III is as in II, but uses 20 cm temperatures. % differences are calculated as: (predicted-observed)/observed.

Figure 1. Daily emergence pattern of DBM adults in each of three trials under four microclimatic conditions (as indicated) conducted from August to November 2000 in Brisbane

Table 2. Mean plant maximum and minimum temperatures (standard deviation) recorded for each treatment run and the degree days (DD) elapsed to the maximum emergence of DBM adults. Values in brackets are the minimum and maximum DD for adult emergence. The predicted emergence time is 285DD.

Discussion

It is well known that the rate of development of an insect is temperature dependent and the relationship is linear above a threshold temperature for a wide range of temperatures (e.g. Allsopp et al. 1991). Having appropriate threshold temperatures, slopes and using the temperatures experienced are important when modelling development time. Predictions were up to 87% higher than the observed development time when default parameter values were used. Default values of threshold temperatures across all immature and adult stages were based on Goodwin (1976). Compared with previously published reports on development of DBM these default threshold temperatures were found to be high. Adjustment brought the predicted development period closer to the observed (Table 1). Use of precise lower threshold temperature values for immature stages of DBM is important to obtain reliable predictions especially when temperatures are fluctuating in this range.

Temperatures recorded by standard Stevenson’s screen are often not the same as temperatures experienced by animals. Considering the average height attained by Brassica crops in the field, best predictions were obtained when daily temperatures recorded at 20 cm above the ground level were used to drive development in the model for Forthside. Use of plant temperatures gave very good predictions of development time for Brisbane conditions as well.

The adjustment of threshold temperatures for each of the life stages of DBM was obtained from previously published reports using linear regression analysis. Development occurring above 30°C, which was above the linear range was neglected from these studies and only that occurring along the linear range was analysed (Figure 2). We selectively removed data that were based on low sample sizes, where temperatures were not well recorded or where animals were checked too infrequently to give good estimates of development time.

Figure 2. Simple scatter plot of all the development rate data for DBM eggs, larvae and pupae we could find in the literature indicating the large variation recorded among studies. No attempt has been made to distinguish studies or to selectively remove points.

The scatter in development values among the studies (Figure 2) can be attributed to the influences of multiple factors, including: differences in temperature and humidity conditions and control (constant versus fluctuating temperatures), host plants used in the experiments, source of DBM and frequency of checking. Published records show the development rate to vary when insects were reared on different cruciferous plants including wild hosts of DBM (Wakisaka et al. 1992). Development time when reared on wild host plants were longer than when reared on cultivated varieties.

Umeya and Yamada (1973) reported differences in development rate between three geographically isolated strains of DBM within Japan. When development rates of Japanese and Thailand strains of DBM were compared they showed no significant difference in development time (Sarnthoy et al. 1989). This view is confirmed by comparing the development performance of nine geographically isolated strains from nine different locations within Asia (Shirai 2000). In the later study, the strains were kept under laboratory conditions for a long time and acclimatization to local conditions by the strains was not checked. In both studies, least minimum temperatures used to test development rates were only 17°C (Sarnthoy et al. 1989) and 15°C (Shirai 2000). Inclusion of more low temperature regimes in such experiments would give an indication of the differences in development rate among strains from temperate zones against those from tropical and sub tropical zones.

Recently, Liu et al. (2002) provided a comprehensive study of DBM development under a range of constant and fluctuating conditions, particularly at high and low temperatures. Assuming there are no major geographic variations in development rate parameters, by using the non-linear functions from this study and appropriate temperatures we will be able to model DBM development anywhere. This model needs to be tested.

Acknowledgements

We thank Lionel Hill for providing access to development data collected in Tasmania, Mike Furlong for DBM eggs and Wayne Rochester for help with statistics.

References