The influence of weather on the population dynamics of the rice stink bug and the implications for integrated pest management

The influence of weather on the population dynamics of Oebalus poecilus, the principal insect pest of rice in South and Central America, is examined. Light trapping was used for one year, but was soon discarded in favour of the more reliable sweep netting which was utilised to monitor populations in fields and non-cropped habitats from autumn 1999 to spring 2002. The results indicate that O. poecilus feeds on alternative host grasses on levées surrounding the rice fields. Adult bugs move into the fields in mid-January and mid-August of each year. In fields there are typically four generations per season, which are approximately four weeks apart. Adults move off the crop in April and October. Data presented here indicate that rainfall is critical to the 'off-season' survival of O. poecilus, and high rainfall during April to July and November to January causes increases in O. poecilus populations. This is particularly important given the recent extreme perturbations in the normal bimodal rainfall patterns caused by El Niño and La Niña events. The ability to predict the severity of O. poecilus populations from weekly rainfall data will provide a significant input into the development of IPM programme for O. poecilus.

Keywords: rice stink bug; Pentatomidae; monitoring; rainfall

In the development of pest management strategies a detailed knowledge of the influence of abiotic factors on the biology of pest insects is essential. Weather and climatic conditions are known to significantly affect the population dynamics of insect pests (Kennedy and Storer, [11]). Knowledge of abiotic conditions, such as temperature, daylength, rainfall and relative humidity can be used as important components in forecasting and predicting the severity of insect pest populations (Milford and Dugdale, [15]). In tropical agroecosystems, temperature and daylength tend to be relatively stable throughout the year and therefore humidity and rainfall exert a greater influence over the population dynamics of herbivorous insects (Milford and Dugdale, [15]).

The small rice stink bug Oebalus poecilus (Dallas) (Hemiptera: Pentatomidae) and related pentatomids are the most serious pests of rice in Guyana (Kennard, [10], Rambajan, [22]) and other South and Central American countries (Halteren, [7], Pantoja et al., [17]). Both adults and nymphs feed on the grains and the damage caused affects the yield and quality of the harvested paddy causing significant losses (Rai, [20]). O. poecilus also feed on several alternative host species of grass weed which grow on the levées surrounding rice fields (Kennard, [10]), although survival is known to be lower on these 'off-season' host plants (Nilakhe, [16]). It is widely accepted that adult bugs move into crops from these alternative hosts when the rice is flowering (Albuquerque, [1]).

In Guyana and much of South America (Costa and Link, [2]), O. poecilus control is achieved solely through insecticides with as many as four preventive sprays being applied each season (Ralph and Rivas, [21]). This is costly for the farmer and also has associated ecological and toxicological hazards. Therefore, in Guyana an integrated pest management (IPM) programme is to be adopted by rice farmers. Monitoring and forecasting pest abundance should ideally be an essential component of any IPM system. Sweep netting has been widely used for assessing the abundance of Oebalus spp. in rice fields and determining action thresholds (Jones and Cherry, [9], Gutierrez et al., [6], Pantoja et al., [17]), but whilst sweep netting provides an effective method of monitoring stink bugs when they have moved into the rice crop, there is no method of forecasting potential increases in populations.

Severe outbreaks of rice stink bugs are common and are believed to be strongly influenced by climatic conditions (Gomez-Sousa and Meneses-Carbonell, [5]), although this has never been proven. Studies concerning the population dynamics of Oebalus spp. are rare and those that do exist tend only to investigate this in the cropped environment (e.g. Taylor, [24], Pantoja et al., [17], [18]). The population dynamics and behaviour during the 'off-season' are largely unknown, as O. poecilus are rarely seen during this period.

Here, the results of work in Guyana are reported where we investigate the abundance of O. poecilus over three years in relation to meteorological data. Insect catches from both light trapping and sweep netting in cropped and non-cropped habitats are related to weather conditions in an attempt to establish their importance in causing fluctuations in O. poecilus abundance. The possible implications of this for the IPM of O. poecilus in Guyana are also discussed.

All field sites were located in the experimental and seed production fields of the Rice Research Station, East Coast Demerara, Guyana, South America (6°27′50′′N, 57°45′25′′W). The rice crop was grown under direct-seeded, lowland irrigated conditions, the two principal varieties being long grain indica rice BR444 and F7-10. Two crops are cultivated per year; the 'spring' crop being sown in December and harvested in April and the 'autumn' crop being sown in June and harvested in October. The climate of the region is wet tropical with a bimodal rainfall pattern (average annual rainfall of 2300 mm). Guyana lies within the equatorial trough zone and its weather and climate are influenced primarily by the seasonal shifts of this trough and its associated zone of rain bands called the Inter-Tropical Convergence Zone (ITCZ). A typical year has two wet seasons—the first in May – June and the second in November – January, with a short dry season from February to April and a long dry season from July to October. Temperatures are highly consistent throughout the year due to the stabilising effect of the Atlantic Ocean and the north-east Trade winds, with a mean maximum of 30.9°C and mean minimum of 25.0°C. Given their consistency, temperature was not considered to be a factor in determining insect abundance and distribution. Meteorological data for the present study came from the nearest continuous recording weather station at Cane Grove, East Coast Demerara (6°37′35′′N, 57°55′00′′W).

A single light trap with incandescent bulb lighting was established in a non-cropped area of the Research Station from 6 May 1998 and operated nightly until 31 July 2000. Insect catches were sorted and identified each morning for weekdays and on Monday morning for weekend trapping periods. Data were combined to give weekly trapping periods.

Weekly standardised sweep net samples of all O. poecilus (adults and nymphs) and other entomofauna were taken over six crop seasons from 24 June 1999 until 13 June 2002. All samples were collected at the same time of day (ca 0900 – 1000) from fields and their associated levées on the Research Station. In the first four seasons, four fields and their levées were monitored and the sample size was increased to eight fields and their levées in the latter two seasons. The average field size was approximately 4 ha. The levées were all 2 – 3 m wide earthbanks vegetated with predominantly grassy species e.g. jungle rice, Echinochloa colonum (L.) and barnyard grass, Echinochloa crus-galli (L.), both of which are known alternative hostplants for Oebalus spp. (Nilakhe, [16]). Sampling from within fields commenced 6 weeks after sowing and continued through to harvest (i.e. 14 weeks within-crop sampling period per season), whilst sampling in the levées continued throughout the year. Each sampling bout, in both fields and levées, consisted of exactly 100 consecutive sweeps (180°) with a 30 cm sweep net (1 sweep with each step forward), at least 10 m in to limit edge effects (Douglas, [4]). Samples were bagged and frozen for later counting and identification. Adult O. poecilus were sexed by dissection.

Whilst O. poecilus nymphs were monitored and recorded in fields, they were omitted from statistical analyses because we were primarily interested in the movements of adult insects into fields. As the nymphs are flightless we can safely assume that movement is over relatively short distances, i.e. plant to plant. All catches were converted to mean per 100 sweeps to permit between-season comparisons and then log10 (n + 1) transformed before ANOVA were applied. Post-ANOVA comparisons of means was done with either Fisher's LSD Multiple-Comparison Test or Tukey-Kramer Multiple-Comparison Test (Minitab™ Release 13).

Rainfall patterns during the three-year study were highly unpredictable. Guyana experienced one of the most severe El Niño events from August 1997 – April 1998 and this was followed by a strong La Niña event in April – August 1999. In northern South America, an El Niño event is expressed as a long period of dry weather and La Niña typically by heavy rainfall. There are obvious implications for this on the population dynamics of O. poecilus.

Rainfall is classed according to the Ministry of Agriculture Hydrometeorological Service's Precipitation Classification for Guyana which may be seen in table 1.

Table 1. Weekly rainfall classification in Guyana (modified from the Ministry of Agriculture Hydrometeorological Service)

A total of 176 adult O. poecilus was caught in a light trap operated each night for one year from 17 June 1999 until 14 June 2000. Over the 12 month sampling period there was a mean of 2.50 (SE ± 0.7) bugs caught per week in the autumn season (1999) and a mean of 4.27 (SE ± 1.2) bugs per week in the spring season (2000). Figure 1 illustrates the numbers of O. poecilus caught over the period, indicating peaks in abundance in the latter half of each season after each harvest. For this reason we did not continue with light trapping in subsequent seasons.

Graph: Figure 1. Total weekly numbers of Oebalus poecilus caught in a light trap at the Rice Research Station, Guyana operated each night for one year from 17 June 1999 until 14 June 2000. (a) autumn 1999 and (b) spring 2000.

In weekly sweep net sampling of rice fields and their associated levées over one year (52 weeks) from 23 December 1999 to 14 December 2000, a total of 2038 O. poecilus adults and nymphs were collected. This included 292 males, 338 females and 494 nymphs from fields and 451 males, 443 females and 20 nymphs from levées. The phenological development of O. poecilus is represented in figure 2. There were two movements of bugs into the fields from levées, one in mid-January and again in mid-August. It is possible that this movement may be a combination of a decline in alternative host-plant quality on the levées and panicle emergence and grain-filling in the adjacent rice crop. In fields there were four generations of O. poecilus per season, which were approximately four weeks apart (figure 2). Numbers of males and females were relatively equal in each of the four generations.

Graph: Figure 2. The phenological development of Oebalus poecilus over two rice crop seasons (spring and autumn 2000) in (a) rice fields and (b) levées at the Rice Research Station, Guyana.

In weekly sampling investigations over six cropping seasons (26 weeks each), a total of 5995 adult O. poecilus were sweep-netted from fields and levées (table 2). The abundance of O. poecilus by sex and by season is shown in table 2. The male to female ratio in the fields was 0.95 : 1 and in the levées was 1.00 : 1. Out of all O. poecilus, 80.2% were netted from the levées surrounding rice fields and in all but two seasons (autumn 2000 and spring 2001) considerably more O. poecilus were caught in the levées than in the fields. Figure 3 shows the mean number of bugs swept per season in fields and levées and illustrates that within-field numbers of O. poecilus are relatively consistent between seasons. An ANOVA on log10 (n + 1) transformed O. poecilus numbers in the field revealed that there were significant differences between seasons (ANOVA, F5,78 2.48, p < 0.05) but there were only differences between spring 2000 and autumn 2001 and spring 2002 (Fisher's LSD test, p < 0.05). Seasonality also had a significant effect on the numbers of bugs found in levées (ANOVA, F5,150 2.68, p < 0.05), with significantly more O. poecilus present on the levées in spring 2000 and autumn 2001 (Fisher's LSD test, p < 0.05).

Graph: Figure 3. Mean numbers of Oebalus poecilus swept per season in levées and rice fields at the Rice Research Station, Guyana. An ANOVA revealed significant differences between the means (p < 0.05). Different letters of the same case above bars indicate significant differences between treatments according to Fisher's LSD test. Bars indicate standard errors of the means.

Table 2. Total numbers of adult small rice stink bug, Oebalus poecilus sweep netted from fields and levées over six rice crop seasons at Rice Research Station, Guyana (1999 – 2002)

The mean numbers of O. poecilus per 100 sweeps per week and the weekly rainfall are shown in figure 4a – f. There appear to be patterns in the temporal abundance of O. poecilus with peaks in numbers in the levées typically occurring in January for spring crops and in August for autumn crops. In the rice fields, peaks of abundance occur from February – March in spring crops and August – October for autumn crops, corresponding with the time of flowering of the rice crop. It is clear from this that O. poecilus populations congregate and feed on the levées, prior to movement into the rice fields at the time of panicle emergence and grain-filling. It is also evident from figure 4 that there is some influence of rainfall on O. poecilus populations, as in seasons such as autumn 2000 and spring 2001 a lack of rainfall caused very low populations on the levées.

Graph: Figure 4. Weekly rainfall and the weekly mean number of Oebalus poecilus from rice fields and their adjacent levées over six rice crop seasons (a) autumn 1999 (b) spring 2000 (c) autumn 2000 (d) spring 2001 (e) autumn 2001 and (f) spring 2002.

If data are plotted against rainfall classes (from table 1), then a clear relationship between O. poecilus populations and rainfall emerges. Figure 5 illustrates this effect where there is an increase in O. poecilus numbers in response to rainfall. An ANOVA on log10 (n + 1) transformed O. poecilus numbers found in rice fields revealed that there were no significant differences between bug numbers (ANOVA, F4,79 0.97, p > 0.05 ns) between the rainfall classes. Rainfall did however have a highly significant effect on the numbers of bugs found in levées (ANOVA, F4,151 10.09, p < 0.001), and significantly more bugs were found in periods when the weekly rainfall exceeded the 'moderately wet to wet' class of precipitation (Tukey's test p < 0.05).

Graph: Figure 5. Mean numbers of Oebalus poecilus caught by weekly sweep netting in rice fields and levées at the Rice Research Station, Guyana by rainfall class. VD = very dry, D – MD = dry to moderately dry, MW – W = moderately wet to wet, VW – EEW = very wet to exceedingly wet and ESW = excessively wet (see table 1). An ANOVA revealed significant differences between the means (p < 0.05) for levées only. Different letters above bars indicate significant differences between treatments according to the Tukey test. Bars indicate standard errors of the means.

The temporal population dynamics of O. poecilus in Guyana was characterised by considerable fluctuations, which are attributed to variations in weather conditions. Reduced rainfall during the critical phase during which O. poecilus should be feeding on alternative host plants in non-cropped habitats, appears to be significant.

The use of light traps for monitoring O. poecilus does appear to give some useful information as to the abundance of pentatomid bugs, contrary to the findings of Link and Grazia ([12]), who claimed that light traps were of little use. The results here indicate peaks of highest abundance in October in the autumn season and in May in the spring. It is evident from the data presented herein that peaks of O. poecilus abundance occur at the end of the cropping cycle, presumably when bugs are moving from harvested fields to 'off-season' alternative host grasses or in search of late-planted crops (Albuquerque, [1]). This would therefore serve as a poor monitoring or forecasting tool for O. poecilus as it would be risky to estimate the next season's population on the previous. Light trapping is however widely used for monitoring of many rice pests in Asia especially for planthoppers (e.g. Manimaran and Manickavasagam, [13], Matsumura, [14]) and rice bugs (Rai et al., [19]). The problems of light trapping are well documented, in terms of the effects of weather on trap catches (e.g. Crummay and Atkinson, [3]) and also the effects of lunar cycles (Ito et al., [8]). These external factors must be borne in mind when using data from light trap catches.

In view of these shortfalls we decided to concentrate on sweep netting. Sweep netting is the standard monitoring tool used for Oebalus spp. in many countries (Jones and Cherry, [9], Taylor, [24], Albuquerque, [1], Pantoja et al., [17]) and provides excellent in situ monitoring of stink bug incidence. Data presented here has demonstrated that sweep netting highlights a build-up of O. poecilus population in both levées and fields. We have shown that in 'typical' seasons, there is an initial build-up of O. poecilus in the levées and then the insects move into the field. This is related to feeding on alternative host plants in the non-cropped habitats, although Nilakhe ([16]) found that Oebalus pugnax (F.) females reared on the alternative hosts, Paspalum urvillei Steud. and E. crus-galli had half the fecundity of females reared on rice plants. This perhaps goes some way to explaining the paucity of nymphal O. poecilus found on the levées.

There are also a number of drawbacks to sweep netting as a form of stink bug monitoring. Firstly, sweeping is a time and labour-costly exercise and second, the majority of farmers and extension officers in Guyana do not have access to sweep nets, although they could be made with little cost. The distinct advantage of sweep netting is that it is known that the numbers of bugs per sweep is directly proportional to the number of bugs per panicle (Gutierrez et al., [6]) and therefore it is relatively simple to establish whether the economic injury level of one bug per panicle has been surpassed. For the immediate future this sampling technique will remain the monitoring tool of choice, although it is still important to stress the importance to extension personnel and farmers of employing this technique correctly and spraying only when thresholds have been exceeded.

Numbers of O. poecilus in the levées varied enormously and these variations we believe are due to rainfall patterns and the resulting availability of alternative food-plant hosts during the between-crop seasons. The effect of rainfall on the distributions of O. poecilus have been demonstrated clearly in this work and it is evident that as rainfall increases so do the O. poecilus populations (figure 4). We propose that it is most important to observe weekly rainfall data during the period of pest population development on levées (January for spring crops and in July for autumn crops). Caution should be applied when weekly rainfall exceeds 42 mm (1 – 2 days), 35 mm (3 – 5 days) or 28 mm (6 – 7 days). Studies in Cuba also established that the flight activity of Oebalus insularis Stål increased in those months when there was high rainfall (Gomez-Sousa and Meneses-Carbonell, [5]).

Rainfall patterns in Guyana are now highly unpredictable because of the severity of the recent El Niño/La Niña Southern Oscillation (ENSO). It is characterised by anomalous warming of the tropical Pacific in the warm phase (El Niño) or cooling in the cold phase (La Niña) with consequent shifts in atmospheric pressure and changes in the wind fields. In the warm phase Guyana experiences reduced rainfall while in the cold phase the country experiences increased rainfall. Guyana experienced one of the most severe El Niño events from August 1997 to April 1998, when there was a drought followed by a subsequent severe La Niña event in April – August 1999 when there was heavy rainfall. This of course has significant implications for the population dynamics of rice pests. Anecdotal farmer accounts claimed that O. poecilus populations were extremely high immediately following the El Niño event. Moreover, the La Niña event experienced during autumn 1999 of this project served to extend the window in which O. poecilus populations could feed in non-cropped habitats adjacent to rice fields. In the future, ENSO is likely to continue to impact on rice production in the tropics (Wailes et al., [25]), therefore an understanding of its influence is vital.

In conclusion, whilst sweep netting from fields monitors the numbers of O. poecilus present in the crop at any one time and light trapping records O. poecilus numbers after the rice crop is harvested, the O. poecilus population may be monitored on levées in advance of severe infestations and could therefore be used as an 'early warning device'. Weekly rainfall data can be used in conjunction with numbers of O. poecilus from levées to predict the severity of pest populations. Such information is essential for effective IPM to reduce farmer sustainability by employing targeted rather than calendar insecticide spraying. This information could be readily integrated into an IPM programme, involving improved cultural practices, conserving and/or enhancing biological control agents (Sutherland and Baharally, [23]), and most importantly rational pesticide use. To offset the increasing costs of rice production in Guyana, pest management of O. poecilus and other pests must encompass all or some of these facets to ensure increased sustainability for Guyanese rice farmers.

We would like to thank Jainarine Harripersaud and Satanand Narain for providing technical assistance and the board of directors of the Guyana Rice Development Board for their support. We also thank Dilip Jaigopaul and Lisa Farnum-Ramjoo at the Guyana Hydrometeorological Service for the meteorological data.