MATERIALS AND METHODS

The present study was conducted in 10 different communities of prefecture of Larissa, one of the main cotton-productive areas of Greece. Monitoring of H. armigera lasted four years from 2002-2005. The monitoring system included four locations within each community and three traps in each location, at a distance between traps of 50 m (on the corners of a equilateral triangle). Traps were of the funnel type, a reusable injection-moulded kind of plastic trap, suitable for dusty locations. Pheromone was z-11-hexadecenyl aldehyde 0.36% w/w (Produced by Russell Fine Chemicals Ltd). Vapona (transfluthrin, 0.4% w/w, distributed by Sara Lee Hellas) is a fumigant insecticide used as standard moth-papers. It was applied in the traps for all the four years, but for years 2003 and 2004 double traps were used additionally, without the presence of insecticide (only with pheromone). Three cotton varieties cover over 80% of the cultivated fields: Celia and Carmen (Fibermax-Bayer) and Sandra (Veterin Hellas).

Data from traps recorded fluctuation of insect population within each community and for the whole Prefecture of Larissa. Environmental conditions (rainfall, relative humidity, temperature) were also recorded. Cotton mean production in kg ha^-1 was also calculated from harvesting data within each community. Correlations between cotton mean production and number of male adult insects were performed. Data transformations, ANOVA and correlations were based on the standard procedures described by Snedecor and Cochran (1980).

RESULTS AND DISCUSSION

Table 1 presents environmental data of Larissa for years 2002-2005. Year 2002 was warmer (mean temperature) and relatively wet compared to other years. Year 2005 was also warm but with very low rainfall. Table 2 presents mean cotton production and adult male insects of H. armigera captured in the 10 communities of Larissa, for years 2002-2005.

In years 2002, 2003 and 2004, the presence of the insect might contributed in lower cotton production. Statistically significant differences between the two types of traps (with or without insecticide application) were found, due to greater number of insects trapped when Vapona was used. The presence of the insecticide obviously contributed in higher number of male catches, because insects were affected by the insecticide and thus, they could not fly away.

Figure 1 presents insect populations monitored within the ten different areas (communities) for H. armigera (traps with or without insecticide application) and Pectinophora gossypiella (Stavridis et al., 2008), according means of years 2003-2004. Fluctuation of insect population was different from year to year. Insect population fluctuations showed statistically significant differences and this phenomenon was described by Deligeorgidis et al. (2008).

Table 1:	Environmental data of Larissa for years 2002-2005
Min: Minimum, Max: Maximum, RH: Relative Humidity

Table 2:	Mean cotton production and adult male insects of Helicoverpa armigera in the 10 communities (C1-C10) of Larissa, for years 2002-2005 and monitoring traps without insecticide application or with insecticide application
H: Harvesting data without replications, All interactions (recording periodxyearxlocation) were statistically significant at p<0.05

Local conditions within each area have been proved important for population fluctuations and these results were completely different compared to data concerning years (Table 2). The specific conditions within each area determine the balance between the insect population and level of damage on cotton production. There were areas where the insect showed increased populations (and possibly low populations of its enemies) resulting in low cotton production. Warm years with light rainfall seem not to favor H. armigera activity (or favor natural enemies of the pest), resulting in lower damages (Table 1, 2). The distinct seasonal fluctuations in insect populations (Walker and Cameron, 1990) have provided opportunities to restrict the application of insecticides to periods when they are necessary for control of the pest.

Fig. 1:	Insect populations monitored within the ten different areas (communities) for Helicoverpa armigera (traps with or without insecticide application) and Pectinophora gossypiella (data adopted from Stavridis et al. (2008)), according means of years 2003-2004

Identifying these periods requires knowledge of economic thresholds and widespread pest monitoring systems. An approach for insect population fluctuations was stated by Sequeira and Playford (2002). Within and between-season patterns of Helicoverpa egg and larval abundance on commercial cotton in central Queensland were analysed using crop scouting data for five farms over nine consecutive growing seasons. Within seasons, analysis of mean weekly egg density profiles for individual farms revealed asynchronous oviposition activity among farms for approximately the first 8 weeks of the growing season. Beyond 8 weeks oviposition activity appeared to be well synchronised among farms. Weekly egg density profiles differed significantly between seasons. The ecological significance of oviposition activity and larval survival patterns on cotton were connected to pesticide application and control strategies.

Correlations on data between years (insect populations and cotton production across all communities) revealed that, when cotton production was low, the number of adult male insects of H. armigera captured in pheromone traps was high (r = -0.62 to -0.69). The opposite was found for P. gossypiella (Stavridis et al., 2008), where correlations were positive. The present study showed that H. armigera populations negatively affected cotton production, requiring monitoring and subsequently, insect population control below damage levels. Statistically significant correlations were not found between cotton production and number of trapped insects when data from all communities (across the four years) were used.

Although correlation data must be handled with care because of the particular biological phenomenon of interaction of insect populations with year conditions (Deligeorgidis et al., 2008) leading in a misfit of insect catches and damage levels and in agreement to the findings of Stavridis et al. (2008) studying populations of P. gossypiella, damage predictions maybe performed from data between years (insect populations and cotton production across all communities). These predictions will refer to the average of damages that will be expected in all communities studied for specific years. The models proposed here, derived from correlations of present study and data adopted from Stavridis et al. (2008). For H. armigera proposed model is hyperbolic y = 3E+42x^-11.415 (r² = 0.65) and for P. gossypiella, model is linear y = 0.0192x-65.462 (r² = 0.87), after best line fit test. Cotton production (y) and subsequently damage levels depend on the total male catches (x), but data are also depended on year conditions estimated by recordings from all communities. Modeling and estimations of Helicoverpa damages on various crops were attempted previously (Chilcutt et al., 2003; Nibouche et al., 2007). Chilcutt et al. (2003) referred that high H. zea (Boddie) densities had no effect on cotton yield unless they occurred late in the season. In particular, this was true for artificial H. zea damage. They also stated that, with the exception of late in the season, the model for simulating H. zea damage to cotton through removal of fruiting structures resulted in similar yields as real H. zea larvae damage to cotton. According to the model by Nibouche et al. (2007), the relative organ (food) availability in the field and the larval instar were found to be significant factors. Chaturvedi (2007) described how monitoring system and preliminary studies of egg and larval distributions and damage assessment trials could be used to develop an initial estimate of a threshold that would maintain damage below the 5% commercially acceptable level for pest H. armigera on Cicer arietinum.

Concluding, correlations on data between years revealed that, when cotton production was low, the number of adult male insects of H. armigera captured in pheromone traps was high. Present study showed that H. armigera populations negatively affected cotton production, requiring monitoring and subsequently, insect population control below damage levels. There were not found statistically significant correlations between cotton production and number of trapped insects when data from all communities (across the four years) were used. Prediction models and practices proposed must be performed with care, because of the year-to-year interactions. We propose only to be used for average estimations within great areas that include data from many locations.

ACKNOWLEDGMENTS

The monitoring program was supported by the Greek Ministry of Agriculture Development and Food. Authors are grateful to the team of agronomists that supported present study, especially to the Head of the Department of Crop Protection Mr. Aristidis Ioannou for his personal interest on this project.