Modeling population dynamics of a tea pest with temperature-dependent development: predicting emergence timing and potential damage
- First Online:
- Cite this article as:
- Satake, A., Ohgushi, T., Urano, S. et al. Ecol Res (2006) 21: 107. doi:10.1007/s11284-005-0099-9
The tea leaf roller, Caloptilia theivora Walsingham (Lepidoptera: Gracillariinae), is one of the serious pests of tea plants in Japan. To understand the mechanism of seasonal occurrence of this insect pest, we developed a population dynamics model that explicitly incorporates the temperature-dependent development of the pest. The model predictions were compared with observed captures in pheromone traps at the experimental site of the Kagoshima Tea Experiment Research Station in Japan. The results showed that the emergence timing of the insect pest observed in the field was determined primarily by temperature. The relationship between the timing of adult emergence and the leaf damage level was also studied using a logistic regression model. The infestation level decreased as the interval between the adult peak emergence date and the date of tea plucking increased, implying that asynchrony between plant phenology and emergence of the insect pest is a critical factor reducing damage level. We examined how the damage level changes according to global warming. Increased temperature made the timing of insect appearance forward and enhance asynchrony of plant–pest phenology. Therefore, reduction of damage level by the insect pest is expected under global warming.
KeywordsTea pestTemperature-dependent developmentPopulation dynamicsSynchrony of plant–pest phenologyGlobal warming
The tea leaf roller, Caloptilia theivora Walsingham (Lepidoptera: Gracillariinae), is a serious pest of tea plants in Japan (Minamikawa and Ueda 1996). This insect pest largely deteriorates the quality of manufactured tea. Leaf damage is caused by the fourth and fifth instars, which make triangular-shaped leaf shelters, contaminating young leaves (Kodomari 1975). There is significant deterioration of both the flavor and taste of tea made with infested leaves, which causes a serious economic problem for the tea industry in Japan.
From the mid-20th century, in order to understand the population dynamics of this insect pest, an annual emergence pattern of adult insects has been monitored by light and/or pheromone traps at several experimental sites widely distributed in Japan (Kagoshima Plant Protection Association 1994). These monitoring programs have provided details of the phenology of C. theivora, but underlying mechanisms that regulate the seasonal occurrence of this species and the relationship between the leaf damage level and adult trap catch remain largely unknown.
Knowledge of the phenology of C. theivora is likely to allow the prediction of the timing of appearance of each developmental stage, which can provide a well-timed control method. A key factor regulating the life history pattern of insect pests is temperature. Because insects are cold blooded, the developmental rates of their life stages are strongly dependent on temperature. In general, the cooler the temperatures, the slower the developmental rate of insects, and as the temperature increases so does the developmental rate until the temperature increases sufficiently to inhibit developmental processes (Gilbert and Raworth 1996). Based on these observations, day-degree (or heat unit) models that assume a linear relationship between temperature and development have been widely used to estimate insect development (Taylor 1981; Pruess 1983; Higley et al. 1986). The nonlinear relationships have also been applied to represent convex features of temperature-dependent development of insects (Stinner et al. 1975; Logan et al. 1976; Sharpe and DeMichele 1977).
Timing of appearance of the insect pest varies, depending on differences in temperature throughout the years, which makes the pest’s forecasting and management difficult. One way to promote our understanding of the phenology of C. theivora is to develop a population dynamics model that explicitly incorporates temperature-dependent development. Indeed models for temperature-dependent development of insect pests have been widely used as decision-support tools to improve the efficiency of pest management, such as accurate forecasting (Graf et al. 1999; Milonas et al. 2001; Yonow et al. 2004) and effective applications of pesticides and viral pathogen (Inoue and Ohgushi 1976; Martin 2001). In addition, there is increased awareness of computerized decision-making tools that incorporate models of insect phenology [DESSAC (Brooks 1998) and MORPH (Walton 1998)].
This study represents a comprehensive attempt to model C. theivora population dynamics using field temperature data. We built a mathematical model in which individuals are categorized according to their life-cycle stages, and the developmental transition is simply described in relation to temperature. The model predicts the seasonal occurrences of all life stages of the insect. We compared the prediction with the observation at the study site of the Kagoshima Tea Experiment Research Station in Chiran, Kagoshima, Japan. Based on the results of the model and data analysis, we predict a possible alteration of damage by the insect pest under global warming.
Materials and methods
The tea leaf roller overwinters as pupa and emerges in spring. Adults emerging from overwintering pupae constitute an overwintering flight to oviposit on the underside of young leaves of a tea plant (Minamikawa and Ueda 1996). Caterpillars hatch, bore into the young leaves, and start feeding. Fourth and fifth instar larvae make triangular shaped leaf shelters by rolling young leaves. The larvae further grow to pupate, and adults emerging from these pupae constitute the next generations. The number of generations per year ranges from four to seven, depending on site-specific climatic conditions, especially temperature (Kodomari 1976).
This insect pest seriously reduces the quality of manufactured tea due to a frass contamination of larvae in leaf nests. If harvest of tea leaves shows a 3% mix rate of contaminated leaves, significant damage is recognized. The leaf damage by first, second, and third instars, which are in the leaf mining stage, rarely influence a growth and yield of tea plant. Significant decrease in tea quality is caused by the fourth and fifth instars in the leaf rolling stage (Kodomari 1975).
Data description and analysis
Daily samples of the insect pest populations were collected from the tea garden owned by the Kagoshima Tea Experiment Station in Chiran, Kagoshima, Japan, using pheromone traps during the period March 1996 to November 2003. Monitoring is still ongoing but we used the data from these 8 years for analysis. At the study site, one pheromone trap was settled, covering a plot of 27 acres in a plantation field of a tea plant race “Yabukita.” Yabukita is the most popular cultivar used in the tea production in Japan because of its high yielding and cold resistance (Hakamada 2003). There were 28 missing values from daily data over the 8 years. These missing spaces were linearly interpolated.
In addition to the pheromone trap catch data, the daily maximum and minimum temperatures were automatically monitored at the study plot using agricultural meteorological equipment (NASCON-2000). These temperature data are used to estimate the developmental rate of each life-cycle stage of the insect pest, as explained later.
Infestation level by larvae was monitored three times per year. The survey was conducted closely according to the tea plucking season. Infestation level was checked by counting the number of infested buds among all buds produced by tea plants at eight plots of size of 50×25 cm during the period of 1996–2003. The Ministry of Agriculture, Forestry and Fisheries of Japan designed the census procedure. These plots were located within the study site where the pheromone trap catch census was took place. To investigate the relationship between the timing of adult emergence and the infestation level, we used a logistic regression model and examined how the infestation level is a function of the interval between the peak date of adult emergence and the date of tea plucking. The peak date of adult emergence was estimated from the pheromone trap data. We considered that the tea plucking dates are consistent with the dates when infestation level was monitored. We used the number of days between the two occasions as an independent variable for the logistic regression. This analysis was applied only to the data from the second tea plucking season that indicate a sufficiently high level of infestation, as described later.
Developmental zeros (°C)
From egg to larva
From larva to pupa
From pupa to adult
Cumulative temperatures (°C)
From egg to larva
From larva to pupa
From pupa to adult
Daily survival rates
Estimated from Matsuhira (2001)
Daily fecundity of female adults
Estimated from Matsuhira (2001)
The parameter values used for simulation runs are shown in Table 1. There are no data available for daily survival rates of eggs, larvae, and pupae. Therefore, we prepared a set of parameters with which we examined how sensitively our conclusions depend on our choice of the parameter values (Table 1).
The major target of our simulation study was to generate within-year dynamics of the pest insect with a given pattern of initial emergence, rather than to predict the among-years dynamics. We adopted such a simulation strategy because (1) no accurate information is available on the mechanism constituting overwintering generations and (2) practically the most important point is to predict the peak occurrence of first and second generations that have the potential to cause economical damage on tea production (Kodomari 1976). Toward that end, we gave independent initial condition for each simulation—the daily number of trapped individuals from March to April in each observed year was used as an overwintering generation that initiates each simulation run. Then simulations were implemented until the flight of the second generation was over, i.e., in mid July.
Observed pest population dynamics
Peak emergence date is a good measure to compare the predicted and observed dynamics of the pest population, because it is very robust to a change of unknown demographic parameters (survival rate of egg, larva, and pupa) while population size is considerably influenced by slight changes in these parameter values (data are not shown). Thus, we verified the model prediction by comparing the predicted and observed peak emergence dates of the first and the second generations.
Using the expected shift on emergence timing (Fig. 8a) and the logistic regression model that estimates infestation level (Fig. 5), we predicted the possible change of damage level by the insect pest under global warming. We assume that plant phenology is independent of temperature. Plant phenology may vary depending on temperature, but the temperature-dependent change of the phenology of Yabukita remains largely unknown. Therefore, we fixed the tea plucking date even under global warming and then calculated the intervals between the estimated peak dates of adult emergence and the dates of tea plucking. The infestation level decreased gradually as temperature increased (Fig. 8b). A 3°C rise of temperature resulted in an almost 3% decline in damage, which was nearly close to the damage-free criterion (i.e., 3% infestation criterion). This decline of infestation level would be explained by asynchrony between the timing of larval hatch and that of leaf flush as explained later.
The long-term pheromone trap data of the tea pest (C. theivora) monitored in Chiran, Kagoshima, Japan, demonstrated that the timing of peak appearance of adult populations varied considerably among years—the maximum time differences of peak emergence dates were as large as 3 weeks. The population dynamics model of C. theivora that explicitly describes a temperature-dependent development clearly showed that the emergence timing of the insect pest observed in the field is determined primarily by temperature, although the potential impacts of other factors, e.g., sensitivity to photoperiod and precipitation, still remain.
Thus, we concluded that by measuring temperature, managers of tea gardens could roughly predict the emergence timing of each of the developmental stages of C. theivora. Knowledge of when each of the developmental stages is present in the field should allow the prediction of when the infestation level will be highest. The logistic regression model for infestation level demonstrated that the infestation level was high (almost two times as large as the damage-free criterion) when the adult population had a peak about 18 days earlier than the date of tea plucking (Fig. 5). The infestation level decreased as the interval between the peak date of adult emergence and date of tea plucking increased, and finally it became sufficiently small as to meet the damage-free criterion (Fig. 5). This can be interpreted by considering the extent to which plant–pest phenology synchronizes. In general, it is pointed out that synchrony of larval hatching and bud burst of host plants strongly affects larval survival (Komatsu and Akimoto 1995; Buse and Good 1996; Dougen et al. 1997). For example, leaf toughness that rapidly changes after bud burst significantly affects the establishment of young C. theivora larvae, hence the timing of oviposition has important fitness implications for this species. Oviposition of the first generation generally coincides with the production of new leaves of host plants. However, when adult moths emerge earlier than foliage flush season, offspring survival must be reduced, which results in low infestation levels. New leaf production of the tea plant normally occurs about 20 days earlier than the tea plucking season on average (K. Uchimura, personal communication). Therefore, the setting of safety criteria—the appearance of adult moths at least 28 days earlier than the tea plucking date (Fig. 5)—has a biologically realistic implication.
Global warming may change the population dynamics of pest species in various ways (Cammell and Knight 1992; Kareiva et al. 1993). For example, an increase in temperature may enhance the overwintering survival and thus increase the number of generations per year (Yamamura and Kiritani 1998), and shift the timing of insect appearance. Logan and Powell (2001) and Powell and Logan (2005) developed temperature-dependent models for insect phenology and demonstrated that global warming may alter adaptive seasonality of many cold-blooded organisms, as exemplified in the mountain pine beetle. In this study, we examined how the damage levels were altered by global warming. Temperature increases up to 3°C made the timing of insect appearance forward and enhance asynchrony between plant and pest phenology, which consequently led to the reduction of leaf damage (Fig. 8). This scenario can be applicable only to the situation in which host plant phenology is mostly independent of temperature. If the relationship between the plant phenology and temperature is investigated in detail, the assumption of the fixed phenology would be relaxed, and then more practical predictions for the damage level would be possible under global warming.
We conjecture that the large differences between the predicted and observed peak emergence dates in 1999 and 2003 (Fig. 3) were caused not only by additional environmental factors such as precipitation and photoperiod but also by man-made events such as pesticide spraying, which greatly alters the population size and potentially changes the emergence timing of the pest. Thus, an analysis incorporating the timing and the impact of pesticide spraying is needed to evaluate the model more accurately.
Several questions remain to be answered in this study. Although our study mainly focused on the emergence timing of the insect pest, population size at the time is of course an equally important measure to determine the infestation level. However, because of little information on demographic parameters, e.g., survival rate of egg, larva, and pupa, we could not analyze the relationship between infestation level and population size. Tea plucking and pesticide spraying that are carried out several times per year may influence the population dynamics of the pest insect, but there are no quantitative assessments of the impact of such man-made operations on the pest’s population dynamics. Therefore, an accurate estimation of unknown demographic parameters and the analysis incorporating the impact of man-made events will be important for the continued development of useful models.
This work was supported in part by a fellowship and a grant-in-aid from the Japan Society for the Promotion of Science (AS), the Ministry of Education, Culture, Sports, Science, Technology grant-in-aid for Scientific Research (A-15207003) (TO), and the 21st Century COE Program (A14) (TO). The authors thank Higuchi S, Iwasa Y, Matsumura M, Nakashizuka T, Shigesada N, Takasu F, and two anonymous reviewers for their helpful comments.