Multiple Insect Pests of Rice - Damage Functions

International Journal of Pest Management Vol. 57, No. 2, April-June 2011, 93–116

Effect of multiple infestations from insect pests and other stresses on irrigated rice in the Philippines: I. Damage functions
J.A. Litsingera*, J.P. Bandongb and B.L. Canapib
a

1365 Jacobs Place, Dixon, CA 95620, USA; bInternational Rice Research Institute, DAPO Box 7777, Metro Manila, Philippines

(Received 4 September 2010; final version received 5 October 2010) Artificial infestation methods were employed to generate damage function graphs for guilds of three rice insect pests. In the vegetative stage, infestation of whorl maggot (Hydrellia philippina Ferino) and defoliators (a mixed population of Naranga aenescens Moore and Rivula atimeta [Swinhoe]) was applied as a combined or single pest attack. ´ Infestation of yellow stemborer (Scirpophaga incertulas [Walker]) and leaffolder (Cnaphalocrocis medinalis [Guenee]) each were infested in one or two crop growth stages. Combinations of three stresses (four N application rates, solar radiation from the wet or dry season, and/or the presence or absence of weeds) were applied to each guild to develop more holistic economic injury levels (EILs). Slopes on the linear portion of damage curves of each pest guild became steeper with each additional stress applied, whether from multiple insect pest attack, low N, low solar radiation, or weeds. EILs can be developed by pest managers from the graphs of the damage functions for the variables quantified in this study. Keywords: Rice insect pests; yield loss; damage functions; multiple pest infestation; weeds; nitrogen fertilisation; solar radiation; artificial pest infestation; damage compensation

1.

Introduction

The mathematical relationship between insect pest injury and yield is called the damage curve or damage function (Pedigo et al. 1986). It is not only the most fundamental, but also the most challenging variable to obtain in the economic injury level (EIL) formula that is used as a basis for farmers’ insecticide-application decision making (Poston et al. 1983; Onstad 1987). The EIL is the lowest population density that will cause sufficient economic damage to justify the cost of corrective control measures. The damage function is more useful to pest managers if the model covers a wide range of infestation levels for determining both the EIL and the economic threshold (Pedigo et al. 1986). The EIL is the slope of the regression equation that defines the damage function. The first attempts to determine damage functions for the most common tropical rice insect pests were conducted in the greenhouse by Dyck et al. (1981). Litsinger et al. (2005) developed an empirical, farmers’ field method where, in each wet and dry season over a number of years, several threshold values were tested side by side in multi-location, replicated trials to determine the most suitable values for the most common chronic insect pests. These trials, carried out from 1979 to 1991, compared the test thresholds to an untreated control and the farmers’ practice side by side with treatments that estimated yield loss needed to evaluate the results. Damage functions could not be

adequately determined due to low natural infestation levels, so the performance of the thresholds was analysed using economic and yield loss criteria after each season and modified as needed in an iterative process of deriving what were termed action thresholds (Litsinger et al. 2006a, 2006b, 2006c). Action thresholds are distinguished from the more rigorous economic thresholds, when the EIL cannot be determined (Hull et al. 1985). The rice crop in the Philippines is most commonly attacked by three main guilds of pests. The green semilooper (Naranga aenescens [Moore]) and green hairy caterpillar (Rivula atimeta [Swinhoe]), that we term ‘defoliators’, along with the rice whorl maggot (Hydrellia philippina Ferino) are the most commonly occurring insect pests in the vegetative stage. In some locations whorl maggot is the sole pest while in others defoliators dominate, but generally they occur as a mixture. As the damage caused by both defoliator species is identical and the larvae are essentially of the same size (see silhouettes of both larvae in Reissig et al. 1986), the two defoliator species are functionally interchangeable. The results showed that, whereas a trained eye could discern the damage, farmers do not make the distinction between whorl maggot and defoliators (Litsinger et al. 2009). Therefore it was deemed more practical to lump them as a single guild and measure their incidence as percentage of damaged leaves regardless of the causal pest. The second and

*Corresponding author. Email: [email protected]
ISSN 0967-0874 print/ISSN 1366-5863 online Ó 2011 Taylor & Francis DOI: 10.1080/09670874.2010.530355 http://www.informaworld.com

94

J.A. Litsinger et al. controlled conditions of experimental stations, the results will not adequately represent on-farm conditions. Savary et al. (1994) working in Central Luzon, Philippines attempted to capture natural site variations by measuring some 30-crop production variables that affect yield. Using correspondence and cluster analyses they found that stemborers were the only insect guild to be associated with high yield loss. Weeds were categorised as occurring above or below the rice canopy, and when in association with stemborers they were one of the main contributors to low yields. In a follow-on study in India, Savary et al. (1997) also showed that, in the presence of weeds, increasing insect pest abundance was more likely to produce linear damage functions showing a lack of compensation; by contrast, with less stress, quadratic damage functions (which showed compensation) were common. This two-part study (see Litsinger et al. Part II 2011) reports on the results of artificial infestation of individual hills in farmers’ fields to generate damage functions for the most common chronic insect pests of irrigated, double-crop rice in the Philippines. Kenmore et al. (1984) had concluded that the rice crop could avoid significant losses if solar radiation were adequate. Litsinger (1993) showed that loss could be alleviated with high N rates. Savary et al. (1994) found that the combination of stemborer infested and weedy fields led to particularly low rice yields. We report how damage functions changed under more realistic field conditions of single or dual insect pest or growth stage infestation under conditions of low and high solar radiation, weedy versus weeded fields, and differing N application rates. 2. Materials and Methods 2.1. Site descriptions and crop management Field trials were carried out in Central Luzon in Nueva Ecija province, the main irrigated rice bowl of the Philippines. A house in Zaragoza was rented that served as a local office, and all trials were carried out on nearby farmers’ fields. A resident on-site research team of the International Rice Research Institute (IRRI), based in Los Banos, conducted the trials. ˜ Zaragoza is located at the tail end of the gravity-fed Upper Pampanga River Irrigation System. The monsoon rainfall pattern consisted of a wet season followed by an essentially rain-free dry season. Further site description can be found in Litsinger et al. (2005). Trials involved IR64, a modern rice variety developed at IRRI with a maturity of 112 d. Transplanting was done at 6–8 seedlings per hill removed from wet seedbeds 4 weeks after sowing. Spacing was 20 6 20 cm between hills. Inorganic N was applied in amounts specified by treatments in two splits, half basal (soil incorporated during the last harrowing before transplanting) and half as a top dressing 7 days before panicle initiation. Also 30 kg P/ha was applied basally to all plots.

third guilds are stemborers and leaffolders. The yellow stemborer Scirpophaga incertulas (Walker) is virtually the only stemborer species recovered from the Central Luzon rice bowl where the current study takes place. ´ Cnaphalocrocis medinalis (Guenee) is one of the key species of leaffolders. Both of these pests are found in greatest abundance during both the reproductive and ripening growth stages. Pedigo et al. (1986) noted that studies to derive EILs in both greenhouse and field locations had drawbacks. Greenhouse studies are poor proxies of farmers’ fields due to an inherently artificial and closed environment as well as lack of inter-plant competition. For example, our previous study using 100-m2 scale field plots replicated 6–8 times and involving natural infestations over a number of years and sites in the Philippines, failed to reveal statistically significant damage functions due to the low pest densities (Litsinger et al. 1987). Thus, if higher densities occur in the future, the model may not be appropriate. Several methods have been used to overcome the aforementioned shortcomings. The first is to use natural infestations but reduce the sampling unit to single hills of transplanted rice plants, as has been done in a number of trials. Shepard et al. (1990) tagged hills that ranged from 0 to 100% damaged leaves from whorl maggot at 3 weeks after transplanting and measured the yield from each hill. Ishikura (1967) and Gomez and Bernardo (1974) expanded the range of damage from stemborers to attain 460% deadhearts or whiteheads. They concluded that to reduce variation, damage functions could be most meaningful by taking samples from the same field and expanding sample size to 100 hills. Pedigo et al. (1986) described various possible mathematical models for damage functions in detail, but for rice stemborers, regressions have been found to follow linear, exponential, or quadratic models using single hills (Ishikura 1967). Artificial infestation is another method used to ensure a wider range of pest infestation levels (Litsinger 2009). Artificially infesting insects has been shown to provide more realistic results rather than simulating damage (Capinera and Roltsch 1980; Rice et al. 1982). Researchers have artificially infested small plots to obtain a wide range of damage levels: Pantoja et al. (1986) used 20 plants/m2, while IRRI (1983, 1984) used 36 hills/2.25m2. Methods of artificial infestation in the field have been developed for whorl maggot (Viajante and Heinrichs 1986), Rivula (Viajante and Heinrichs 1987), and stemborers (Soejitno 1977; Dang 1986; Bandong and Litsinger 2005). Damage functions from the literature have been derived mostly for single pests, but in most fields, crops are stressed by not only other insect pests but also factors such as weeds, plant diseases, weather, plant nutrition, variety, cultural practices (Seth et al. 1969; Turnipseed and Kogan 1987). It thus follows that if damage function studies are performed under the

International Journal of Pest Management 2.2. Field trials Four of the most common rice insect pests were compared either as individual or combined species or as individual species with infestations in one or two rice growth stages. Three different experiments were replicated in the 1989 wet and 1990 dry seasons to accommodate three pest guilds. The first experiment involved both whorl maggot and Naranga and Rivula defoliators, each infesting in separate treatments and combined as a third treatment. These three species occur together in the rice vegetative stage, thus damage functions need to account for their joint damage. The other two most common rice pests occur in the mid and late growth stages. They most commonly are abundant as single pests but will damage two successive growth stages. Thus, in the second experiment with the yellow stemborer, the first treatment involved infestation of the reproductive growth stage, while the second involved the ripening stage. The third treatment was the combined infestation of both growth stages. The third experiment involved leaffolder, where in the first treatment infestation occurred in the reproductive stage, with the second in the ripening stage, and the third occurred in both. A further variable for each experiment included sub-treatments of four levels of N fertiliser (0, 30, 60, and 90 kg N/ha). Finally, all treatments were divided into plots where weeds were controlled (weed-free) and conversely where weeds were allowed to grow (weedy). Each replication of the set of three experiments per season comprised different farmers’ fields, each planted over a 2–3-week period so as to spread the workload in the randomised complete block design. Insect pests were loaded onto 25 randomly selected individual hills per replicate covering an area of 1 m2. Each experiment was replicated in six farmers’ fields for a total of 150 hills per treatment per season. Fields were established very early in each season to avoid bias from natural insect pest infestation. Artificial infestation was employed in order to obtain a wider range of damage levels than naturally occur. There were two infestation phases per growth stage for each pest. The first phase for whorl maggot and defoliators occurred on 20 of the 25 hills per plot leaving five hills uninfested. Field collected whorl maggot eggs are rarely parasitised (Jahn et al. 2007), so mortality comes largely from predators. However, predation is generally low at the beginning of each season. Leaf sections, each containing single eggs, were cut from plants in nearby fields. At 7 days after transplanting (DT) (first phase), half of the 20 hills were each infested by two eggs and half each with four eggs. Leaf sections were affixed with household white glue. During the second phase at 10 DT, half of each set of hills was infested with four eggs so that sets of five hills had 0, 2, 4, 6, or 8 eggs/hill. Maps were made of each replicate to keep track of which hills had been infested; hills were

95

selected randomly. In treatment 2, second to third instars of mixed Naranga and Rivula larvae were obtained from rearing colonies in the IRRI Entomology Department. The same infestation arrangement was followed as with whorl maggot. Five hills were uninfested and ten hills each received one larva while ten hills received two larvae. On 10 DT, half of the hills infested at 7 DT with one or two larvae, received an additional two larvae so that sets of five hills each had 0, 1, 2, 3 or 4 larvae/hill. The vegetative stage extends to 35 DT. Stemborer and leaffolder were infested in each of two plant growth stages: (1) prior to panicle initiation (reproductive stage) 35 DT (first phase) and 42 DT (second phase) and (2) at panicle exsertion (ripening stage) 56 DT (first phase) and 63 DT (second phase). Leaffolder second to third instar larvae were obtained from IRRI headquarters from a rearing colony of C. medinalis. Four larvae were placed on 10 hills and 8 others were placed on 10 other hills with 5 hills uninfested. In the second phase, 5 hills that had received 4 or 8 larvae received an additional 4, while 5 hills that received 8 larvae received an additional 8 so that each set of 5 hills bore 0, 4, 8, 12, or 16 larvae/hill. The infestation rate was higher than for defoliators as the crop was older and hence each hill bore many more leaves. Yellow stemborer moths, collected at outdoor lights, were caged on potted rice plants. Medium-sized egg masses were selected for rearing in Petri dishes and inspected at the blackhead stage following Bandong and Litsinger (2005). A cage was placed for one week to exclude natural enemies from contacting larvae before the larvae entered tillers. As Scirpophaga egg masses occur in varying sizes, egg masses were cut with a razor to a size of a prescribed circle that would encompass on average about 50 eggs. Masses free of egg parasitoids were placed at three per 1 m2 (25 hills), either 35 DT for the reproductive stage or 63 DT for the ripening stage. The dual stage infestation totalled 6 egg masses, 3 in each stage. Insect damage was assessed on all 25 hills per 1-m2 plot for all treatments. In the whorl maggot and defoliator experiment, the percentage of damaged leaves was determined 35 DT when all leaves were scored as either damaged or undamaged. Damage assessment from reproductive stage infestation of leaffolder was carried out at 56 DT and the ripening stage infestation at the flag leaf stage 77 DT following the same manner. Reproductive stage stemborer infestation was assessed as percentage of deadhearts and whiteheads at panicle exsertion, while ripening-stage infestation was assessed at 10 days before harvest. Hills were hand harvested with grains oven dried to 14% moisture and weighed on a top-loading electronic balance accurate to +0.1 g. In each treatment, the damage function was determined by ranking infestation levels per hill from lowest to highest for each

96

J.A. Litsinger et al. 3. Results All weeds compete for nutrients, space, and water, but those that grow above the crop canopy also compete for solar radiation. The main weeds encountered below the canopy in the six damage function trials in Zaragoza were Monochoria vaginalis (Burm.f.) Presl., Cyperus diformis L., Paspalum distichum L., and Scirpus supinus L. Species above the canopy were dominated by Echinochloa spp., followed by Scirpus maritimus L. and Ischaemum rugosum Salisb. The 144 regressions covering two seasons were graphed as six figures (Figures 1–6) with their equations given in Tables 1–3. The regressions showed highest fitness in only two of the five mathematical models tested. The linear regression model was most common, having negative slopes that showed lack of compensation where the rate of yield loss was constant over all infestation levels. Less common was the quadratic model that indicated crop compensation to have occurred, particularly at lowest infestation levels (Pedigo et al. 1986). We found that, within each pest guild, insect injury was overcome more when infestation occurred as a single pest or in a single growth stage, with plants fertilised at higher N levels, in the dry season (more solar radiation) or in weeded plots. Plant stress was associated with steep slopes in less differentiated regression equations and tended to increase with maximum yield loss and decreasing N rates. The details of these findings are now presented. 3.1. Whorl maggot and defoliators: individual and combined infestation 3.1.1. Whorl maggot The artificial augmentation of rice whorl maggot and Naranga and Rivula defoliators achieved a range of densities that were higher than normally occur in the field for both pests, attaining 50% or more damaged leaves when infested individually and 80% or more when combined. Mean infestation levels for the wet and dry seasons for the complex of whorl maggot and defoliators were 2–3.5 times more than the action threshold of 410% damaged leaves, respectively (Litsinger et al. 2006a). Graphical damage functions for whorl maggot as a single species infestation are presented in Figure 1A, B for the wet season and Figure 2A, B for the dry season. Figure 1A represents the damage function without weeds, while Figure 1B shows the interaction of increasing densities of whorl maggot under weed stress on yield measured in grams per hill. For each N level, as whorl maggot infestation increased, yield decreased as described in the mathematical model. Within each season or weed condition, with some exceptions, yields rose progressively higher with each increase in N level. In comparing the figures we see that some are more differentiated than others. Lower differentiation between the regressions in Figure 1B indicates less yield

replicate. As all data were scored as a percentage, all hills within a class interval of 5% damaged leaves or deadhearts/whiteheads were averaged in terms of mean yield per hill in order to generate the regressions depicted in Figures 1–6. In each class, means were computed only when 42 entries occurred. In weeded plots all the weeds were totally removed by hand pulling. The crops were stressed further by varying N levels. The effect of season is mainly due to differences in the amount of solar radiation received by the crop. An actinograph set out in the agro-met station in Guimba in 1989–1990, nearby to Zaragoza, recorded an average of 28% more solar radiation in the dry season (430 calories/cm2/d) than wet season (310 calories/cm2/d). Yoshida (1981) confident that, in a monsoon climate, particularly at a northern latitude such as the Philippines with short days, the wet season rice crop would be stressed by lack of solar radiation. These criteria were employed to assess the degree of plant stress associated with the treatments: (1) the proportion of regressions that were quadratic versus linear, (2) maximum yield loss, and (3) mean slope of regression equations. Maximum yield loss was calculated from each damage function for each N level by subtracting the yield (g/hill) under the highest infestation level from the uninfested condition (0% level from the x-axis on each graph) divided by the uninfested yield and multiplied by 100. Another variable included averaging the slopes of each set of regressions for the X variable. Steeper slopes would suggest the plants were under greater stress. A ranking system was used for mean slope and maximum yield loss. Mean slopes were separated at increments of 0.500 from 0 or positive slopes to 72.00. Low stress was indicated from values ranging from positive down to 0 which showed no stress, 0.0001 to 70.5000 (low), 70.5001 to 71.000 (moderate), 71.001 to 71.5000 (high), and 471.5001 (very high). Maximum yield losses were also ranked: 0–25% (low), 26–50% (moderate), 51– 75% (high), and 475% (very high). Plant stress was further indicated from the more regressions that were quadratic and how well the regressions were differentiated. Clumped and nondifferentiated regressions were considered as stressed. As a means of assisting in the evaluation of the results we adopted a rating system of plant response to the four combinations (quadrants) of weediness and solar radiation in terms the being most to least stressful based on the aforementioned criteria. 2.3. Statistical analysis

The data were subjected to ANOVA regression analysis from each treatment using SAS software. The best fitting mathematical model (highest r value) was selected based on greatest statistical significance. In the software package, the models tested for best fit were linear, exponential, logarithmic, power, and quadratic.

International Journal of Pest Management

Figure 1. Damage functions of rice whorl maggot and defoliators under four rates of inorganic N ( 90 kg, 60 kg, competition. Insects were artificially infested as single pests and combined. Zaragoza, Nueva Ecija, Philippines, 1989 wet season.

30 kg,

0 kg N/ha) with and without weed

97

98 J.A. Litsinger et al.

Figure 2. Damage functions of rice whorl maggot and defoliators under four rates of inorganic N ( 90 kg, 60 kg, competition. Insects were artificially infested as single pests and combined. Zaragoza, Nueva Ecija, Philippines, 1990 dry season.

30 kg,

0 kg N/ha) with and without weed

International Journal of Pest Management 99

Figure 3. Damage functions of stemborer under four rates of inorganic N ( 90 kg, 60 kg, 30 kg, 0 kg N/ha) with and without weed competition. Stemborers were artificially infested during the reproductive and/or the ripening stages of rice. Zaragoza, Nueva Ecija, Philippines, 1989 wet season.

100 J.A. Litsinger et al.

Figure 4. Damage functions of stemborer under four rates of inorganic N ( 90 kg, 60 kg, 30 kg, 0 kg N/ha) with and without weed competition. Stemborers were artificially infested during the reproductive and/or the ripening stages of rice. Zaragoza, Nueva Ecija, Philippines, 1990 dry season.

International Journal of Pest Management 101

Figure 5. Damage functions of leaffolder under four rates of inorganic N ( 90 kg, 60 kg, 30 kg, 0 kg N/ha) with and without weed competition. Leaffolders were artificially infested during the reproductive and/or the ripening stages of rice. Zaragoza, Nueva Ecija, Philippines, 1989 wet season.

102 J.A. Litsinger et al.

Figure 6. Damage functions of leaffolder under four rates of inorganic N ( 90 kg, 60 kg, 30 kg, 0 kg N/ha) with and without weed competition. Leaffolders were artificially infested during the reproductive and/or the ripening stages of rice. Zaragoza, Nueva Ecija, Philippines, 1990 dry season.

Table 1. Summary of the regression equations for whorl maggot þ defoliators damage functions from infesting different rice growth stages based on artificial infestation of individual rice hills. Crops were further stressed by allowing weeds to grow as well as from decreasing N rates. Seasonal differences relate to the amount of solar radiation for crop growth. Slopes were derived from the regression equations which were either linear (no compensation) or quadratic (compensation). Zaragoza, Nueva Ecija, Philippines 1989 wet season and 1990 dry season. 1/. Wet season Weedy P 6.78 0.004 0.590 5.99 0.01 6.92 0.001 5.06 0.02 5.21 0.002 0.310 Y ¼ 26.2 þ 0.061X 7 0.041X2 0.847 0.894 0.836 Model R
2

Dry season No weeds F P Model R
2

No weeds F P Model R2 F

Weedy P 7.34 7.20 6.64 0.009 0.002 0.008 0.820 10.34 50.0001

N level

Model

R

2

F

0.851 11.78

0.779

9.62

Whorl maggot 0 Y ¼ 21.03 7 1.142X 30 Y ¼ 20.94 7 0.967X 60 Y ¼ 20.60 7 0.755X Y ¼ 18.9 þ 0.434X 7 0.081X2 Y ¼ 18.9 þ 0.601X 7 0.083X2 Y ¼ 22.7 þ 0.152X 7 0.062X2 0.001 71.093 Y ¼ 20.78 7 1.24X 0.724 Y ¼ 10.1 7 0.687X 0.748 8.26 0.002 Y ¼ 12.5 7 0.867X 0.811 12.01 0.0003 Y ¼ 17.6 7 0.589X 7 0.068X2 0.584 7.23 0.006 Y ¼ 15.1 7 0.219X 7 0.057X2 70.593

0.805

7.77

50.0001 Y ¼ 18.58 0.797 7 1.45X 50.0001 Y ¼ 15.20 0.723 7 0.870X 50.0001 Y ¼ 15.78 0.590 7 0.831X

90

Y ¼ 21.86 þ 0.365X 7 0.027X2 Slope mean ¼ 70.634 0.006 6.58 0.005 7.38 0.01 0.004 Y ¼ 27.8 7 0.026X 7 0.082X2 Y ¼ 29.4 þ 0.229X 7 0.095X2 8.88 0.002 0.884 13.45 0.005

0.801

0.801

0.784 10.41

0.785 0.863 0.872

5.76 6.89 9.93 0.882 12.03

0.03 0.004 0.0002 0.0001

Defoliators 0 Y ¼ 17.6 7 0.674X 30 Y ¼ 18.5 7 0.643X Y ¼ 13.4 0.564 7 0.695X 0.0002 Y ¼ 16.8 0.766 7 1.10X 0.0005 Y ¼ 16.8 0.602 7 0.849X 0.003 70.905 Y ¼ 21.0 0.818 10.2 7 0.965X Y ¼ 20.3 - 0.303X 7 0.046X2 Y ¼ 25.1 7 0.562X 7 0.051X2

0.807

6.34

60

0.807

7.19

90

Y ¼ 19.3 þ 0.863X 7 0.063X2 Y ¼ 20.5 þ 0.449X 7 0.904X2 Slope mean ¼ 71.387 0.03 0.809 0.783 0.844 0.832 8.04 0.003

0.802

5.82

Y ¼ 15.0 7 1.01X 0.923 11.01 0.0001 Y ¼ 18.7 7 1.14X 7 0.022X2 0.823 9.09 0.006 Y ¼ 20.4 7 0.028X 7 0.120X2 0.777 5.49 0.009 Y ¼ 23.8 7 0.418X 7 0.073X2 70.650

Y ¼ 19.1 7 0.767X

0.765

7.32 0.005

0.668 0.462 0.631

6.92 3.01 5.70

0.003 0.04 0.006

7.30 0.0001 Y ¼ 21.0 7 0.736X 6.42 0.005 7.77 0.006

International Journal of Pest Management

Whorl maggot þ defoliators 0 Y ¼ 19.8 0.808 5.02 7 0.966X 30 Y ¼ 20.7 0.804 8.98 7 0.945X 60 Y ¼ 20.7 0.827 13.96 7 0.721X 90 Y ¼ 22.0 0.875 6.22 7 0.795X Slope mean ¼ 70.867

Y ¼ 15.5 7 0.838X 0.005 Y ¼ 17.8 7 0.916X 50.0001 Y ¼ 18.3 7 0.874X 0.007 Y ¼ 17.8 7 0.754X 70.845

Y ¼ 25.9 7 0.400X 7 0.030X2 Y ¼ 25.6 7 0.136X 7 0.056X2 70.513

Y ¼ 8.08 7 0.362X 0.800 7.44 0.001 Y ¼ 9.81 7 0.339X 0.940 13.56 0.002 Y ¼ 10.3 7 0.372X 0.875 12.82 0.0002 Y ¼ 15.1 7 0.720X 70.448

0.859

4.99

0.008

103

1

/ Slopes averaged from the X values.

104

Table 2. Summary of the regression equations for stemborer damage functions from infesting different rice growth stages based on artificial infestation of individual rice hills. Crops were further stressed by allowing weeds to grow as well as decreasing N rates. Seasonal differences relate to the amount of solar radiation for crop growth. Slopes were derived from the regression equations which were either linear (no compensation) or quadratic (compensation). Zaragoza, Nueva Ecija, Philippines 1989 wet season and 1990 dry season. 1/. Wet season Weedy P 0.926 0.828 50.0001 0.005 0.0003 0.703 0.809 5.79 0.0001 0.789 9.53 14.34 0.0009 0.824 7.34 6.49 0.002 0.840 6.90 14.28 0.0002 0.854 12.37 0.009 0.730 0.782 0.854 0.850 70.783 Model R
2

Dry season No weeds F P Model R
2

No weeds F P Model R2 F 6.08 6.90 13.21 8.88

Weedy P 0.0006 0.0002 0.0001 0.0003

J.A. Litsinger et al.

N level Y ¼ 19.2 7 1.994X Y ¼ 18.3 7 1.392X Y ¼ 20.0 7 1.886X Y ¼ 20.5 7 1.805X 71.765 Y ¼ 17.7 7 1.00X Y ¼ 20.1 7 1.124X Y ¼ 21.9 7 0.222X 7 0.091X2 Y ¼ 24.3 7 0.500X 7 0.065X2 70.710 Y ¼ 15.7 7 1.114X Y ¼ 14.5 7 0.892X Y ¼ 15.9 7 0.253X 7 0.656X2 Y ¼ 16.9 7 0.880X

Model

R

2

F

Stemborer reproductive stage infestation 0 Y ¼ 19.1 0.845 7.99 0.0003 7 1.332X 30 Y ¼ 20.0 0.955 10.05 0.0004 7 1.227 X 0.901 13.82 0.0003 60 Y ¼ 20.5 7 0.662X 7 0.086X2 0.896 9.37 50.0001 90 Y ¼ 19.6 7 0.478X 7 0.174X2 Slope mean ¼ 70.923 0.905 0.874 0.650 0.768 6.77 0.0001 3.12 0.04 0.626 0.836 5.89 0.0004 0.488 11.12 0.0005 0.484 4.03 7.39 10.45 15.52 0.02 0.0003 50.0001 0.0004

0.04

0.475 0.616 0.472 0.860

5.11 8.90 5.33 14.53

0.02 0.0001 0.005 0.0008

Stemborer ripening stage infestation 0 Y ¼ 19.9 0.478 3.11 7 1.118X 30 Y ¼ 19.7 0.541 3.41 7 1.175X Y ¼ 18.9 7 1.395X Y ¼ 19.4 7 1.669X Y ¼ 13.6 7 0.043X Y ¼ 17.2 7 1.139X 71.055 0.741 0.748 0.708 0.774 7.09 15.74 9.35 17.83 50.0001 0.0006 0.004 0.0003

0.04

60

Y ¼ 23.2 7 1.583X

0.788

8.88

0.005

Y ¼ 8.93 7 0.506X Y ¼ 10.7 7 0.368X 7 0.026X2 Y ¼ 11.7 7 0.463X Y ¼ 15.9 7 0.696X 7 0.035X2 70.508 0.536 0.585 0.677 0.488 5.09 6.66 8.43 10.33 0.006 0.0007 0.002 0.001

90

Y ¼ 22.7 7 1.366X

0.751

11.12

0.0003

Y ¼ 16.1 7 0.72X Y ¼ 15.5 7 0.423X 7 0.025X2 Y ¼ 14.2 7 0.749X 7 0.111X2 Y ¼ 20.2 7 0.245X 7 0.048X2 70.535

0.536 0.585 0.677 Y ¼ 12.4 7 0.630X 70.558 0.488

5.09 6.66 8.43 10.33

0.006 0.0007 0.002 0.001

Stemborer reproductive þ ripening stage infestation 0 Y ¼ 16.8 0.887 14.56 0.0009 Y ¼ 13.1 7 0.918X 7 0.718X 30 Y ¼ 17.3 0.859 7.82 0.0002 Y ¼ 12.2 7 0.886X 7 0.505X 60 Y ¼ 16.4 0.706 6.03 0.008 Y ¼ 14.0 7 0.771X 7 0.668X Y ¼ 15.4 7 0.741X 70.660

Y ¼ 9.91 7 0.479X Y ¼ 11.2 7 0.516X Y ¼ 12.4 7 0.595X

90

Y ¼ 17.9 7 0.809X

0.751

8.52

0.001

Y ¼ 13.2 7 0.138X Y ¼ 13.3 7 0.118X Y ¼ 14.4 7 0.019X 7 0.001X2 Y ¼ 14.4 7 0.019X 7 0.001X2 70.105

1

/ Slopes averaged from the X values.

Table 3. Summary of the regression equations for leaffolder damage functions from infesting different rice growth stages based on artificial infestation of individual rice hills. Crops were further stressed by allowing weeds to grow as well as from decreasing N rates. Seasonal differences relate to the amount of solar radiation for crop growth. Slopes were derived from the regression equations which were either linear (no compensation) or quadratic (compensation). Zaragoza, Nueva Ecija, Philippines 1989 wet season and 1990 dry season. 1/. Wet season Weedy P 0.777 0.833 0.709 0.672 8.42 7.40 6.28 0.002 8.49 0.002 Model R
2

Dry season No weeds F P Model R
2

No weeds F P Model R2 F

Weedy P 9.66 0.0002 6.55 0.0009 8.44 0.0002 6.88 0.0003

N level Y ¼ 12.6 0.785 7 0.913X Y ¼ 15.2 0.870 7 1.128X

Model

R

2

F

Leaffolder reproductive stage infestation 0 Y ¼ 19.1 0.856 6.33 0.004 Y ¼ 14.7 7 1.165X 7 0.915X 30 Y ¼ 21.4 0.719 5.40 0.0003 Y ¼ 15.5 7 1.143X 7 0.913X

60

0.0005 Y ¼ 18.1 0.867 7 1.396X 0.0003 Y ¼ 18.3 0.696 7 1.121X

90

0.441 4.79 0.0004 Y ¼ 17.1 Y ¼ 17.6 7 0.971X 7 0.513X 7 0.104X2 0.755 6.45 0.0005 Y ¼ 17.7 Y ¼ 17.9 7 1.100X 7 0.950X 7 0.143X2 Slope mean ¼ 70.633 71.093 0.899 13.55 0.854 9.77

0.822 7.61 0.0001 Y ¼ 16.3 7 0.9841X 0.757 9.03 0.0006 Y ¼ 20.0 7 0.534X 7 0.067X2 0.827 8.61 0.0008 Y ¼ 21.2 7 0.011X 7 0.135X2 0.879 5.99 0.0002 Y ¼ 19.6 þ 0.769X 7 0.135X2 0.310

Leaffolder ripening stage infestation 0 Y ¼ 18.7 0.836 5.03 0.0003 Y ¼ 18.2 7 1.210X 7 1.363X

0.0004 Y ¼ 10.9 0.495 7 0.576X 0.0003 Y ¼ 12.3 0.532 7 0.568X

6.55 0.002 7.48 0.007

30

60

0.798 10.03 50.0001 Y ¼ 16.4 0.740 10.66 0.0005 7 0.882X 0.919 13.88 5.0001 Y ¼ 16.0 0.716 7 0.750X 70.695 7.57 0.0003

90

0.624 5.61 0.0004 Y ¼ 18.2 Y ¼ 18.3 7 1.282X 7 0.432X 7 0.057X2 0.762 7.40 0.003 Y ¼ 17.3 Y ¼ 16.3 þ 7 1.100X 1.400X 7 0.208 X2 0.719 9.18 0.0005 Y ¼ 18.1 Y ¼ 22.0 þ 7 1.071X 0.444X 2 7 0.128X Slope mean ¼ 70.053 71.203 0.624 4.67 0.003

0.760 5.49 0.0003 Y ¼ 19.1 7 0.387X 7 0.075X2 0.656 5.69 0.0002 Y ¼ 19.5 7 0.107X 7 0.776X2 0.792 6.02 0.0007 Y ¼ 19.0 þ 0.379X 7 0.107X2 0.871 7.34 0.005 Y ¼ 21.2 þ 0.406X 7 0.140X2

0.880

7.53

0.004

9.50 0.0001 7.44 0.0004 8.44 0.009 0.857 19.44 50.0001 Y ¼ 14.9 0.785 7 0.685X 70.840 5.21 0.006

Leaffolder reproductive þ ripening stage infestation 0 Y ¼ 19.4 0.936 4.88 0.002 Y ¼ 12.9 7 1.091X 7 0.623X 30 Y ¼ 19.2 0.887 6.35 0.0003 Y ¼ 16.0 7 1.094X 7 1.091X 60 Y ¼ 19.3 0.750 4.98 0.003 Y ¼ 14.8 7 0.939X 7 0.886X

Y ¼ 11.6 0.806 7 0.869X 0.908 13.96 50.0001 Y ¼ 14.9 0.756 7 0.779X 0.879 18.77 50.0001 Y ¼ 18.3 0.800 7 1.021X

International Journal of Pest Management

90

Y ¼ 16.5 7 0.775X 0.583 5.55 0.004 Y ¼ 21.2 7 0.786X 0.735 7.45 0.002 Y ¼ 24.7 þ 1.031X 7 0.013X2 0.893 6.34 0.0001 Y ¼ 16.77 0.293X 0.925 7.44 0.0008 Y ¼ 24.0 Y ¼ 24.3 7 0.768X 7 0.067X2 þ 0.660X 7 0.024X2 7 0.031X2 Slope mean ¼ 70.973 70.723 0.030

105

1

/ Slopes averaged from the X values.

106

J.A. Litsinger et al. mean slope and maximum yield loss were high, and there was only a quadratic regression at the highest N rate. Also, the linear regressions were non-differentiated for the other three N rates. The next most stressful quadrant was the weed-stressed plots in the dry season; the mean slope indicated a high level of stress as did the maximum yield loss. Also, there were only two quadratic regressions representing the highest two N rates that were able to compensate. The third most stressful quadrant was that of the weed-free, wetseason crop; stress rankings were both moderate for slope and maximum yield loss. Only the 90 kg N regression was quadratic. The least stress occurred in the quadrant which was a weed-free, dry-season crop which had a positive slope and low yield loss. 3.1.2. Defoliators The damage functions for defoliator injury in weedfree plots in the wet season (Figure 1C) resulted in quadratic equations for the two highest N levels, showing the greatest tolerance. The degree of compensation was high, and did not precipitously decline until 20–25% damaged leaves was reached. The slopes of the two linear regressions for 0 and 30 kg N/ha were highly similar (70.674 and 70.593) (Table 1) and moderately steep, but were 24% lower yielding than the higher N treatments. The mean slope over the four N levels was a high 71.387 and the maximum yield loss averaged a moderate 36% (range 30–46%). Yield levels of the four damage functions were well differentiated showing the benefit of higher rates of N that overcame stress. In the dry season, all four N rates exhibited quadratic regressions (Figure 2C). The shapes of the two regressions for 60 and 90 kg N/ha indicated compensation up to 15–20% damaged leaves while those for 0 and 30 kg N/ha showed minimum compensatory capacity. As in the wet season, yield levels were distinctly higher for each increase in N rate. The mean slope was a low 70.165 and maximum yield loss was a moderate 34% (range 29–39%). In the wet season under weed pressure (Figure 1D), all N levels had damage functions that were linear showing the greater stress under the combination of weeds and wet season weather. The mean slope of the regressions was a moderate 70.905. The damage functions were of similar moderately steep slopes, but yield was 29% higher in the 90 kg N/ha plot over the range of damage levels than the other three N rates. Although the linear regression of 60 kg N showed a moderate response above the two lower rates, maximum yield loss was a moderate 54% (range 45–63%). In the dry season despite the weed stress, the 30, 60 and 90 kg N/ha treatments revealed quadratic well differentiated equations (Figure 2D). The 0 kg N/ha regression therefore showed the steepest slope. The mean slope was a moderate 70.650. Maximum yield

response to N than occurred in Figure 1A and was probably due to the suppressive effect of weeds. In the wet season without weed stress, the quadratic regression model was the best fit only at 90 kg N/ha (Figure 1A), but yield was consistently higher over all infestation levels than the 0–60 kg N/ha rates showing the benefit of greater crop nutrition in tolerating injury. The quadratic damage function, however, did not deviate greatly from a linear equation. With decreasing N levels, the slopes of the regressions became more steeply negative (at 60 kg N/ha the slope was 70.0967 while at 0 kg N/ha was 71.142) (Table 1) indicating increasingly higher rates of yield loss as leaf damage rose. The average slope of all four regressions was 70.634 which is considered moderate. Progressively higher stress, measured as maximum yield loss, occurred as N decreased from 60 to 0 kg N/ha: 39%, 47%, and 58% loss. For example at 0 kg N/ha yield decreased from 19.9 g/hill at 0% damaged leaves to 8.4 kg N/ha at 50% damaged leaves resulting in a 58% yield loss while at 60 kg N/ha yield declined from 20.0 g/hill to 12.2 or 39% loss. With the maximum loss at 90 kg N/ha of 29%, the overall average was 43% which is considered moderate. The dry season weed-free condition (Figure 2A) gave a different result as all four N levels resulted in quadratic equations and yield loss was not significant until damaged reached 20–25% damaged leaves. The four regressions representing increasing increments of N were well differentiated and yield rose progressively with each higher rate with the greatest yield response at 90 kg N/ha. This rate showed the greatest tolerance to whorl maggot injury and the regression was 16% higher than the 60 N/ha rate. The mean slopes of the damage functions were less steep in the dry season (þ0.310) (Table 1) which suggests the crop responded positively to the higher solar radiation. Maximum yield loss averaged a low 22% (range 15–25%) in the dry season which was half that of the wet season. In the wet season (Figure 1B) no compensation occurred at any N level in the weedy plots as all damage functions were linear. The 0 kg N/ha rate had the steepest slope (71.45) (Table 1) which registered an 85% maximum loss while the other N rates averaged 67% (range 57–68%). In the dry season (Figure 2B), damage functions were quadratic in the two highest N rates in weedy plots; even so, the slopes were still steep and degree of compensation was slight. The average slopes (Table 1) were steeper in the wet season (71.093) versus 70.593 in the dry season in the presence of weeds. Maximum yield loss, however, was a high 66% (range 51–80%), equal to that of the wet season. Weeds produced high stress as can be seen just by comparing the yield levels between weeded and weedy plots which were 27 and 22% lower, respectively for wet and dry seasons. In summary, whorl maggot injury was made most severe in the weedy, wet-season quadrant. Here both

International Journal of Pest Management losses were higher than for the weed-free treatments with a high mean 69% (range 55–78%). In comparing the four quadrants, the most stress occurred in the weedy, wet-season weedy plots where regressions had very steep mean slopes and moderate maximum yield loss. There were only linear regressions but the 60 and 90 kg N treatments showed a yield advantage and compensation. There were two quadrants that tied for the next highest level of stress. The first was also in the wet season without weed stress, while the second was the weedy, dry-season weedy. Both revealed very steep mean slopes and moderate maximum yield losses. Two regressions were quadratic. The quadrant with the least stress was the dry-season, weed-free plot where all regressions were quadratic and the mean slope was less steep while maximum yield loss was moderate. 3.1.3. Whorl maggot þ defoliators

107

Only linear regression equations were noted for the combined whorl maggot þ defoliator infestation during the wet season both with (Figure 1E) and without (Figure 1F) weed stress. The average moderate slopes were nearly identical (70.867 and 70.845) (Table 1), although yield levels were notably lower in weedy treatments. This illustrates the difficulty the rice crop faces with multiple stresses from whorl maggot þ defoliators plus weeds under the lower solar radiation in a wet season. Maximum yield losses were high 68% (range 58–79%) in weeded, but a very high 78% (range 69–93%) under weed stress. In both figures, the regressions were highly non-differentiated thus there was little yield advantage from using higher N rates. Weeds caused a moderate 40% yield loss in the wet season. In the dry season in weed-free conditions (Figure 2E), the 60 and 90 kg N/ha rates responded similarly with nearly identical quadratic damage functions, while the 0 and 30 kg N/ha rates had similar linear functions but 38% lower yields. Notably all the functions were quadratic (Figure 2A, C and E) whether either pest was infested singly or multiply, at the two highest N levels. The damage functions for each N level in the wet season (Figure 1E) were more clumped than those of the dry season (Figure 2E). The 90 kg N/ha treatment showed a markedly steeper slope (70.720) compared to the other three N rates (70.339 to 70.372) (Table 1). Maximum yield losses were similar among the four N rates ranging from 57% (range 54–62%) in weeded to 64% (range 47–75%) under weed stress, but were notably lower than in the wet season under weed-free (68%, range 58–79%) and weedy (78%, range 69– 93%) treatments. In summary, the highest stress was again in the weedy-wet season quadrant which had very high maximum yield and all the functions were quadratic and a moderate regression slope. Regressions were all linear and undifferentiated. The next stressful quadrant

was again in the wet season but in weed-free plots. Again all regressions were linear and undifferentiated, but maximum loss was high while the mean slope was moderate. The third most stressful quadrant was in the dry season in association with weeds. While again all regressions were linear, the 90 kg N/ha rate did show a distinct yield advantage indicating some compensatory response had occurred. The steepness of the mean slope was rated low while the maximum yield loss was high. Least stressful was the dry season, weed-free plot which had the two highest N rates exhibiting quadratic equations. All regressions were well differentiated and maximum yield loss was high, but the mean slope was only moderately steep. We would expect that the combination of whorl maggot, an internal feeder, and the external feeding defoliators, would produce additional stress on the crop when both were present in the same rice growth stage. This appeared to be true for whorl maggot þ defoliators. An examination of the wet-season, weedfree plots (Figure 1A, C, E) in particular shows higher stress from the dual infestation as all the regressions were linear compared to only 2–3 with single pest infestation. Also, dual infestation revealed less differentiated regressions with steeper slopes and higher maximum yield loss. In weedy, wet-season plots (Figure 1B, D, F), regressions in the dual infestation were less differentiated and maximum yield loss was higher. Turning to the dry season without weed stress (Figure 2A, C, E), the dual infestation produced fewer quadratic regressions which had steeper slopes and higher maximum yield losses. Under weed stress (Figure 2B, D, F) there were again fewer quadratic regressions and overall yield level was lower than in the single pest infestations. 3.2. Stemborer single and dual crop stage infestations

Introduced yellow stemborer egg masses produced 35– 55% deadhearts and whiteheads when infested in a single stage and from 70 to 75% when combined. Mean infestation levels for both seasons for stemborer injury in the reproductive stage were above the action threshold of 15% deadhearts which was just exceeded for both the wet (19%) and dry (26%), but in the ripening stage the result reached 18 and 30%, 1.8 and 3 times the 10% deadheart threshold (Litsinger et al. 2006c), respectively. 3.2.1. Stemborer: reproductive stage

Stemborer injury during the reproductive rice growth stage in weed-free conditions in both the wet and dry seasons (Figures 3A and 4A) produced similar damage functions over each N rate. Yield levels were well differentiated in both seasons. Also the 60 and 90 kg N/ha treatments produced quadratic regressions of similar shape in both seasons, while the lower N rates

108

J.A. Litsinger et al. In the dry season, all but the 0 kg N/ha treatment produced quadratic equations in weeded plots, whereas under weed stress both the 30 and 90 kg N/ha plots did so (Figure 4C, D). The most striking contrast between weeded and weedy conditions occurred with the 60 kg N/ha treatment. When weeds were removed there was compensation up to a level of 20% damaged leaves, whereas no compensation occurred under weed stress. Maximum yield losses were high averaging 53% (range 49–55%) when weeded and 60% (range 42–73%) when not. Yield benefit from weeding was 38% over all N rates but the slopes were moderate and similar: 70.535 (weeded) versus 70.508 (weedy) (Table 2). In summary, the greatest stress once more occurred in the weedy, wet-season quadrant where mean slopes and maximum yield loss were high and all regressions were linear with undifferentiated N responses. The second most stressful quadrant this time was again in the wet season but weed-free. Once more only undifferentiated linear regressions were evident and the mean slope and maximum loss were high. This was followed by the quadrant where the two highest N levels produced quadratic functions and maximum yield loss was high but mean slope was only moderate. All regressions, however, showed yield increases with increasing N levels. Least stressful was the weed-free dry season with three quadratic equations showing compensation with moderate slopes but high maximum yield loss. 3.2.3. Stemborer: reproductive þ ripening stages With stemborer attack in both the reproductive and ripening crop stages (Figure 3E and F), rice plants could not compensate well in the wet season, and the linear damage functions were almost overlapping among all N rates in both the weeded and weedy conditions showing a poor response to N. The mean slopes were moderate and similar: 70.848 (weeded) and 70.660 (weedy). Yields, however, were 28% lower in the weedy plots, whereas maximum yield losses between weeded and weedy plots were similar and bordered on being very high: 73% for weeded (range 66–80%) and 70% for weedy conditions (range 58–82%). In the dry season, infestation without weeds (Figure 4E) produced quadratic regressions for the two highest N rates with notably higher compensation at 90 kg N/ ha. The dry season showed greater capacity for compensation than the same weed-free conditions in the wet season, and mean slopes were less steep (70.105 versus 70.848) (Table 2). The presence of weeds in the dry season produced only linear damage functions (Figure 4F), with notably steeper slopes (70.558) than when weeded (70.105) (Table 2). Maximum losses, however, were similar and high, averaging 73% in the weeded condition (range 64– 80%) and 74% under weed stress (range 68–82%).

resulted in linear regressions. The dry season produced 17% higher yields for the 60 and 90 kg N/ha treatments than in the wet season. The mean slopes were moderately steep in both the weed-free wet season 70.923 versus the dry season 70.710 (Table 2). The high maximum yield loss levels were also remarkably similar, averaging 55% (range 50–61%) in the wet season and 57% (range 54–60%) in the dry season. In the wet season under weedy conditions (Figure 3B), the four N-rate linear damage functions were highly similar in yield level with steep slopes where higher N rates showed little yield benefit particularly at higher infestation levels. Maximum yield loss averaged a high 67% (range 53–78%). In the dry season (Figure 4B), the 60 kg/ha rate under weed stress produced a quadratic curve, but compensation was minimal. Yield levels were 27% higher for the top two N rates. The dry season seemed to improve the ability of the crop to tolerate injury under weed stress as the mean slopes were less steep (70.783 versus 71.765) (Table 2). Maximum yield loss averaged 64% (range 54–73%). Insect pest yield loss combined with weeds averaged 11% in the wet season but 28% in the dry season over the four N rates. In summary, within the reproductive stage infestation, we estimated that the greatest stress occurred in the quadrant with the combination of wet season and weeds as there were no quadratic regressions and both the mean slopes and maximum yield losses were considered high. In addition the regressions were undifferentiated indicating that, due to the high stress higher N rates were not able to generate high yields. Weeds again brought about the second highest stress level, this time in the dry season where the 60 kg N/ha rate produced a quadratic function which overlapped with that of 90 kg N/ha. Maximum yield loss was high and mean slopes were moderate. The third highest stress occurred in the wet season without weed stress. Here two quadratic regressions were produced and both maximum yield loss and mean slopes were moderate. Least stress occurred in the weed-free, dry season where again two quadratic regressions were registered but although the mean slope was moderate, maximum yield loss was low. 3.2.2. Stemborer: ripening stage

There was little difference in the wet season between weeded and weedy treatments for stemborer infestation in the ripening stage for all N levels as all eight regressions were linear (Figure 3C, D). Mean slopes also were similar 71.314 versus 71.055 (Table 2), but the mean yields over the four N levels were 20% higher in the weeded condition. Only the slope of the 60 kg N/ ha treatment in the weedy plots was noticeably less steep than the others. Maximum yield losses were high, averaging 59% (range 53–59%) in the weeded but a more varied 54% (range 29–77%) under weed stress.

International Journal of Pest Management In summary there was a tie among the three quadrants for exhibiting the most plant stress. They were both wet season crops as well as the weedy, dryseason crop that only produced linear equations as well as exhibited high maximum yield losses and moderately steep slopes with undifferentiated N levels. Least stressful was the weed-free dry season with two quadratic regressions and less steep mean slope, although maximum yield loss was high. In comparing single with dual stage infestation, more linear regressions occurred in the latter in weedy, dry-season plots suggesting greater stress. In addition the values for maximum yield loss were higher (73%) in both the wet and dry seasons with and without weed stress compared to single stage infestation losses 58%. However in comparing the yield levels of the four season-by-weed stress combinations, one does not discern any appreciable decrease as a result of dual stage infestation. 3.3. Leaffolder single and dual crop stage infestation

109

Artificial infestation of leaffolder produced up to 50% damaged leaves when infested in separate stages and up to 75% when combined. The average artificial infestation was very similar in each trial for both growth stage infestations (26–29% damaged leaves) and was just more than twice the action threshold level of 15% damaged leaves (Litsinger et al. 2006b). 3.3.1. Leaffolder: reproductive stage The damage functions in the wet season derived from leaffolder larvae infested in the weed-free reproductive stage (Figure 5A) resulted in quadratic regressions at both the 60 and 90 kg N/ha rates. Compensation occurred at damage levels up to 20% damaged leaves for both N levels. Both the 0 and 30 kg N/ha rates had similar very steep linear regression slopes (71.165 and 71.143) (Table 3) with the only difference being the 19% lower yield levels for 0 kg N/ha. Therefore compensation began between the 30 and 60 kg N/ha rates. Under weed stress (Figure 5B), all of the regressions were linear with similar slopes where the only differences were slightly higher yields for each higher N rate. The added stress of weeds on top of stemborer damage increased maximum yield loss from 49% (range of 36–65%) for the four N rates to 59% (range 56–65%). Yield level was 39% less when under weed stress. Also the slopes of the equations were more steep under weed pressure (71.093, highly steep) than the weeded condition (70.633, moderately steep) (Table 3). Therefore the presence of weeds in the wet season repressed compensation from even the highest N rates, and the range of slopes was narrow between all N levels (70.913 to 71.100). Reproductive stage infestation in the dry season without weed stress showed a slightly greater capacity

for the crop to compensate than in the wet season as the three highest N levels resulted in quadratic regressions (Figure 6A). The 90 kg N/ha rate in particular showed similar capacity for compensation as in the wet season. The dry season crop showed a more differentiated response to N which underscores the role it plays in increasing the plants’ ability to tolerate damage. The mean slope of the wet season was twice as steep (70.633) as in the dry season (0.310) (Table 3) to suggest that solar radiation also contributed to greater compensatory ability. Under weed infestation, all damage functions were linear with steep slopes, whereas for the 0–60 kg N/ha rates, the regressions converged at the peak 50% damage level (Figure 6B). The presence of weeds removed the ability of the crop to compensate at any N rate, as slopes were steeper (70.593) than the weed-free condition (0.310). In the dry season without weed pressure, maximum losses over the four N levels averaged a high 57% (range 40–65%) compared to a very high 75% (range 65–81%) with weed stress. This compares to 49% loss in the wet season without weed stress. Thus in this case an exception to the prevailing trend occurred as the stresses of leaffolder damage plus weeds had a greater depressing effect on yield in the dry than wet season. The quadrant with the greatest stress was again a weedy plot but this time in the dry season. Maximum yield loss was very high with a moderate regression slope. Second highest stress came in the weedy, wetseason plot. Mean slope was high as was maximum yield loss. The N rates exhibited in the linear regressions were more differentiated in the dry season but yields were actually lower than in the wet season. In the wet season without weed stress, the higher two N rates produced quadratic equations giving us the third most stressful combination. Maximum yield loss and mean slopes were only moderate, however, and the N-rate regressions were differentiated. The least stressful combination was the weed-free dry season which produced three quadratic equations and even a positive mean slope. Maximum yield loss, however, still was high. 3.3.2. Leaffolder: ripening stage

Ripening stage infestation in the wet season without weed stress (Figure 5C) revealed quadratic regressions for the 60 and 90 kg N/ha rates, while lower N rates were linear. Compensation occurred up to a level of 25% damaged leaves at the 60 kg N/ha rate. There were also well differentiated separations in yield level as N rate increased. Under weed stress (Figure 5D), all of the damage functions were linear, however, the 90 kg N/ha rate showed a notably less steep slope and 32% higher yield than for the other N rates to suggest better compensation. The regressions from the other three N rates were highly similar. Maximum losses from ripening stage injury averaged 55% (range 40–64%) over all N rates in the weed-free condition, but rose to

110

J.A. Litsinger et al. N rates were 32% lower yielding than the quadratic regressions. Not removing weeds (Figure 6F) increased overall stress as now all N rates produced linear regressions that were 32% lower yielding on average than without weeds. Also the mean slope under weed stress was much steeper (70.840) than without weeds (0.030) (Table 3). Maximum yield losses, however, were not different as a mean of 78% (range 73–85%) occurred in weeded and 80% under weed stress (range 73–86%). The greatest stress under leaffolder infestation occurred in the quadrant in the dry season with weed stress. Mean slopes were moderate but maximum yield loss was very high and N rates were undifferentiated. Second greatest stress came as a tie in both the weedy and weed-free wet season where both had only a single quadratic regression at 90 kg N/ha which showed higher compensatory capacity. Maximum yield losses were very high while slopes were moderate. Least stressful was the quadrant in the dry season without weed stress where both 60 and 90 kg regressions were quadratic. Although maximum yield losses were very high, the mean slope was positive. Comparing single stage infestation with dual stage infestation we note that, particularly in the weed-free conditions, the number of quadratic functions relative to linear ones was less under infestation at both stages in the dry season (1 quadratic versus 2–3) and wet season (2 quadratic functions versus 3–4) (Figures 5 and 6). From Table 3 we see that the mean slopes were higher in the multiple stage infestation in the wet season without weed stress as well as in the dry season under weed stress. Overall maximum yield loss was higher in the dual stage infestation 81 versus 59% for single stages. 4. 4.1. Discussion Methods of artificial infestation

65% (range 37–80%) under weed stress. The 90 kg N rate registered the lowest maximum loss (37%). The mean slope of the weed stressed regressions was very much steeper (71.203) than without weeds (70.053) (Table 3). Yield of weeded plots averaged 27% higher than weed-free. In the dry season in the absence of weeds (Figure 6C), all of the damage functions were quadratic which showed compensation from leaffolder damage up to 10–15% damaged leaves, particularly at the 60 and 90 kg N/ha rates. In contrast, the regressions under weed stress (Figure 6D) were all linear and 43% lower yielding. The damage functions for the 60 and 90 kg N/ ha treatments under weed stress overlapped, averaging 26% higher yield than the lower two N rates. The mean slopes of the regressions were much steeper under weed stress (70.695) compared to weeded (0.073) (Table 3). Despite the different slopes, the maximum yield losses were notably similar between the weeded (mean 58%, range 52–67%) and weedy (mean 53%, range 47–57%) treatments. Most stress occurred in the quadrant of unweeded plots in the wet season where both mean slopes were steep and maximum yield loss high. Only the 90 kg N/ ha rate resulted in a significant yield response. All the regressions were linear as was the case in the next most stressful quadrant: the dry season under weed stress. Maximum yield loss was also high but the mean slope was less steep. The third most stressful quadrant was the wet season without weeds which resulted in three quadratic equations and positive mean slopes. Maximum loss was, however, high. The dry season without weeds produced the least stress where all regressions were quadratic, but maximum yield was high. 3.3.3. Leaffolder: reproductive þ ripening stages

Infestation of both the reproductive and ripening growth stages in the wet season without weeds resulted in only one quadratic regression (90 kg N/ha) (Figure 5E). The 0–60 kg N/ha regressions were linear and clumped in terms of slope and yield level, while the 90 kg N/ha rate yielded 33% higher. A similar result occurred under weed stress in the wet season (Figure 5F) where the 90 kg N rate produced the single quadratic equation, while the others were overlapping linear regressions. Thus the mean values of the slopes were similar between weed-free (70.973) versus weedy (70.723) (Table 3), but the yield was 31% higher without weeds. Maximum yield loss of all four N rates, however, was only 5% higher in the presence of weeds (mean 86%, range 72– 100%) than weeded (mean 81%, range 73–89%). In the dry season without weed pressure (Figure 6E), both the 60 and 90 kg N rates produced quadratic regressions. The degree of compensation, however, was not high as slopes were steep. The 60 kg N/ha regression averaged only 6% lower yield than at 90 kg N/ha, while the regressions of other two lower

Our study improved artificial methods for infesting individual rice hills in farmers’ fields for whorl maggot, defoliators, and leaffolder. The method of hand placement of insects on rice plants was labour intensive but provided a wider range of infestation densities than occurs naturally in the field. Hand placement avoids the problem of having to set cages over the crop for extended periods, but it limits solar radiation, and also so confounds results by introducing another crop stress. Viajante and Heinrichs (1986) placed cages over 2.25 m2 field plots of 49 hills and introduced up to 800 field-collected whorl maggot adults which achieved a wide range of damaged leaves. Percentages are unknown, however, as a 1–9 grading scale was used where a value of 9 was 450% damaged leaves. Their purpose, however, was not to derive damage functions. Introducing adults is more rapid than hand placement of eggs but requires caging. The cages were removed

International Journal of Pest Management after 3 weeks. In our trials, whorl maggot eggs were affixed singly on individual leaves up to 8 eggs per hill. The resulting infestation ranged from 0 to 50–60% damaged leaves. In future trials using egg placement, the range should be increased to 16 eggs to hopefully attain 100% damaged leaves. With defoliators, we hand placed up to 4 larvae per hill. The result was that there were ample numbers of hills in most 5%-incremental classes from 0 to 50–55% except the last few. Again we should have doubled the number of infested larvae to attain 100% damaged leaves. Viajante and Heinrichs (1987), using another method, placed 0, 2, 4, 8, 15, or 25 pairs of Rivula moths per 2.25 m2 (49 hills) and achieved 0–89% damaged leaves, but left a gap at the lowest infestation levels between 0 and 25% damaged leaves. Therefore, another treatment of one mated pair per cage would be called for. The moths were caged for 4 weeks. For leaffolders we placed up to 16 larvae per hill and achieved a range of 0–50% damaged leaves for both the reproductive and ripening stages, but some classes at the higher infestation range had no entries. To achieve the 100% defoliation level we should have doubled infestation levels to 32 larvae per hill. No other studies of artificial leaffolder infestation in the field were found in the literature. There are studies in the literature where yellow stemborer egg mass infestation was also employed (e.g. Litsinger 2009). We placed 3 medium sized egg masses with about 50 eggs each per 25 hills (1 m2). The resulting stemborer infestation ranged from 0 to 55% deadhearts/whiteheads. In future trials, 6 egg masses should be used per m2. In the study of IRRI (1983), up to 500 first-instar stemborer larvae were placed on 36 hills in cages, but damage levels were not given. Hand placement of egg masses seems to be the most practical method for artificial infestation of stemborers, particularly when natural enemies are minimised. In the four pest guilds, increasing the number of infested hills above 150 would also have increased the chance of values in all classes from 0 to 100% damage. 4.2. What information the shape of the damage curve conveys Pedigo et al. (1986) recognised six parts of a generalised damage function that plots as a sine curve with the apex being damage-free. The first three parts occur at the lowest infestation levels where the crop is able to overcome much insect injury. Tolerance is the first part and is defined as no damage per unit of injury with a zero slope, while the second is over-compensation which is negative damage defined as a yield increase with a positive slope. The third is compensation, which occurs when there is increasing damage per unit area in a curvi-linear relationship, but with an increasing negative slope along the linear portion. The fourth is linearity which occurs when there is constant

111

maximum damage and a constant negative slope. The two final segments do not always occur, but when they do, they take place at the tail end of the curve as it changes from a linear to a less steep slope. The first is desensitisation which indicates a decreasing slope and the second is inherent impunity which is a zero slope at a level above zero yield. Damage functions in this study produced two models, most were linear with the balance being quadratic. These damage functions resulted in varying degrees of recovery from loss (compensation), but eight (6%, n ¼ 144) showed tendencies of over-compensation at low injury levels based on the mathematical models: defoliators as the sole pest in Figure 1C wet season at 60 and 90 kg N/ha and dry season Figure 2C at 90 kg N/ha, stemborer at the ripening stage in dry season Figure 4C at the 60 kg N/ha rate, and leaffolder at the reproductive stage (Figure 5A) at 90 kg N/ha and at the ripening stage (Figure 5C) at 60 kg N/ha in the wet season as well as in Figure 6A at 90 kg N/ha and Figure 6C at 60 kg N/ha in the dry season. Thus over-compensation at some damage level was implied for all pests except whorl maggot. It is rare for insect injury to lead to a 100% loss, and in our study there were no examples of a zero yield in any damage function, although most regression lines came close. This was because we were not able to achieve infestation levels much above 60% damage levels as single pests or above 90% in mixtures. Therefore we were not able to measure if desensitisation or inherent impunity would have occurred. The majority of damage functions published for rice insects have been with stemborers resulting mostly in linear regressions as the model of best fit (Litsinger 2009). Linear regressions were also recorded with leaffolder (Murugesan and Chelliah 1983) and rice bug Leptocorisa spp. (van Haltern 1979). Studying stemborers in Laguna province, Gomez and Bernardo (1974) found that the exponential model fit best. There were two differences between our studies and that of Gomez and Bernardo that may explain the difference. Aside from the different varieties used (IR22 and C4– 63), the main species in their study was the striped stemborer Chilo suppressalis (Walker) whose larva is about one third larger than Scirpophaga spp. based on head capsule diameter (Rothschild 1971). In addition, 4–6 striped stemborer larvae normally feed in a tiller compared to the usual single Scirpophaga larva (Shiraki 1917), thus the striped stemborer would cause more damage to the plant’s vascular system. The only other studies where the exponential model was found as a best fit was that of Ishikura (1967) in Japan where striped stemborer dominated, although linear and quadratic mathematical models were also common. 4.3. Effect of crop stresses on yield The weedy, wet-season imposed the greatest stress in six of the nine insect pest treatments (all three of the

112

J.A. Litsinger et al. growth stage infestation of stemborer and leaffolder. Part of the explanation for the higher stress of whorl maggot and defoliators, we believe, is due to transplanting shock which impinges early root development. Severe root injury results when farmers pull their seedlings as the heavy clay soils are not conditioned with organic matter. Therefore, farmers delay pulling until the seedlings are sturdy enough to withstand the pressure. Later on, the crop ‘outgrows’ transplanting shock so that in the later growth stages this stress does not contribute to loss from stemborer or leaffolder. Transplanting older seedlings, in turn, reduces tillering and thus yield, further exacerbating the stress (DeDatta 1981). These results support the yield loss trial reported in the companion study. 4.3.2. Weed stress Savary et al. (1997) concluded that the combination of weed stress and insect pest injury was likely to produce linear damage functions. We corroborate these findings and conclude that removal of weeds can greatly enhance the crop’s ability to tolerate insect pest damage during any rice growth stage. The farmer cooperators in our study were good managers so that weed loads were only moderate when not controlled in the treatments. Had weed biomass been greater we would have seen even greater stress levels produced. Weeds compete for the same nutrients needed by rice so their presence greatly impedes growth. If weeds can rise above the rice canopy then very high losses are assured. Weeds, of course, rarely occur as one species in rice fields, and in our study seven species dominated in the two seasons, representing the three main weed groups: grasses, sedges and broad leaves. In effect, weeds occur as multiple species stresses. Farmers know that weeds can severely reduce rice yield, but often underestimate the weed infestation and delay in carrying out good management practices. The main control methods are good land preparation, ponding, application of weedicides, and hand-pulling from the both field and surrounding bunds. Therefore there exists an array of control methods that farmers can use. Farmers, in a hurry to plant, will forego some ploughing or harrowing operations, and at times water supply is not timely so that the required depth of ponding cannot be maintained. Although transplanting is a method developed by farmers to give rice plants an advantage over weeds in field establishment, if ponding is not 5 cm deep, weeds will rapidly emerge and their growth can outpace that of rice. In the study area only 30–50% of farmers used weedicides, therefore it is important that labour be made available for hand-weeding, which is not always possible. The study site is at the tail end of the irrigation system so water delivery can at times be problematic. Hand-weeding must therefore be carried out from the third week after transplanting.

whorl maggot and defoliator combinations and in single plant stage infestations for stemborer and the ripening stage for leaffolder). The remaining three treatments were divided among weedy, dry-season (twice) and weeded, wet-season (once) for most stresses. In the nine treatments, the least stress occurred in the weeded, dry-season. We therefore conclude that weeds and the wet season imposed added stress that greatly reduced the capacity of rice plants to tolerate the imposed insect pest injury. N deficiency was also shown to exacerbate the imposed suite of stresses (weeds, solar radiation deficiency, and multiple pests). We now discuss each of these stresses. 4.3.1. Multiple insect pest/stage stress We conclude that the addition of a second pest in the same growth stage (whorl maggot þ defoliator) or infestation in two growth stages (stemborer and leaffolder) increased stress levels of rice plants. We used five measurements to evaluate the degree of additional stress caused. Whorl maggot and defoliators showed that when both pests were together they produced more linear regressions relative to quadratic models (seven of the eight quadrants), with the only exception being whorl maggot in the weedy, wetseason. The same comparison for stemborer produced five linear regressions and leaffolder six, and these were dominantly in the dry season. Multiple infestations also produced steeper mean slopes in five of the eight comparisons for whorl maggot and defoliators compared to one each for the other two pests. Regarding stemborers, notably steeper slopes occurred in the weedy, dry-season in reproductive stage injury. For leaffolders steeper slopes also occurred only in the reproductive stage injury, but in the weeded, dryseason. Maximum yield loss was higher for whorl maggot and defoliators when combined than when each pest was infested singly in five of the eight comparisons. Loss was higher for leaffolder four times but only twice for stemborer when both growth stages were infested. There was also notably less differentiation among the regressions in five of the eight comparisons for whorl maggot and defoliators compared to three times for stemborer and twice for leaffolder. Less differentiation indicates that the plants were under stress from various causes that dampened their response to applied N. Yield levels were also notably lower in the dual stage infestation of whorl maggot and defoliators in three of eight quadrants, whereas this only occurred twice for stemborer and not at all for leaffolder. The results showed not only increase stress from dual stage infestations for all the chronic pests, but also that this occurred more prominently in whorl maggot þ defoliators than for the other two pests. These findings support the results of Litsinger et al. (2011) where yield loss was found to be synergistic for whorl maggot and defoliators, but additive for dual crop

International Journal of Pest Management 4.3.3. Solar radiation deficiency How different seasons affect crop yield and insect pest abundance is measured by more than just the incidence of solar radiation. In the wet season, heavy rain can affect yield indirectly by knocking off insect pests from rice plants (Nakasuji and Dyck 1984), and high humidity encourages insect pathogens (Rombach et al. 1994). In the dry season high temperature and low humidity increase mortality of eggs and young larvae (Suwongwan and Catling 1987). But we rated hills on the degree of damage, so that the aforementioned effects of weather on insect pest population development is not relevant in this study. The wet season is mainly stressful due to lower solar radiation resulting not only from the cloudy monsoon weather, but also from shorter day lengths particularly during the ripening stage (Yoshida 1981). Kenmore et al. (1984) documented greater expression of rice insect pest damage during cloudy weather than in full sunshine. Wet season yields are typically 20% lower than in the dry season at the latitude of C. Luzon (DeDatta 1981). Farmers have little recourse as to management, even with the knowledge that earlier planting would avoid the peak typhoon period as the irrigation system produces hydro-electricity for Metro Manila and water release is politically based on the latter rather than the needs of farmers. Therefore N recommendations for the wet season are proportionally reduced based on the seasonal yield potentials. Solar radiation deficiency is a key stress in its own right but we have showed that it is more so when in combination with the other stresses tested herein. On the other hand, we have shown that the rice plant is able to compensate from insect injury and other stresses, often to a large degree under dry season solar radiation. 4.3.4. Nitrogen deficiency

113

were those of the 90 kg N/ha rate and the lowest when nil. The exceptions to this result were when the plants were under severe stress from the other causes tested in this study. Usually the regressions were well differentiated in ascending order with increasing N rate. Quadratic regressions, if they occurred, were normally those with the highest N rates. Quadratic regressions for the 0 N/ha rate become the best fit only when stresses were removed, such as in the weeded, dry-season. In the rice plant, increased N application increases average leaf size, number of leaves per shoot, number of shoots per hill, number of grains per panicle, and percentage of filled grains (DeDatta 1981). Each of these variables, directly or indirectly, increases yield. Those factors known to be stimulated by high N rate, such as enhanced appetite (more plant injury) and better nutrition (greater pest survival) probably occurred in this study. The third known factor is higher fecundity which probably did not become relevant due to the small plot size. But none of these factors likely affected the outcome of this study as the damage functions were based on the resulting damage, not population dynamics. However, as far as making recommendations for the farmer, these factors become a consideration. However in double-cropped rice systems, insect pest abundance is significantly reduced by the action of natural enemies particularly where insecticide usage is minimized (Ooi and Shepard 1994). Modern high-yielding, semi-dwarf rices, such as IR64 used in this study, are known for their ability to tiller in response to N application (DeDatta 1981). Tall, traditional rices on the other hand are inherently low tillering and only elongate more in the presence of N which leads to lodging and low yield. To ameliorate pest stimulatory effects on the crop, it is recommended to split N application 2–3 times depending on crop maturity, balance inorganic usage with organic N sources, and not to apply excessive N rates. 4.4. Effect of multiple stresses There were five stress factors in this study: (1) increasing pest densities, (2) more than one insect pest feeding in the same growth stage or the same pest feeding in two growth stages, (3) reduced solar radiation in the wet season, (4) weeds, and (5) N deficiency. We have shown that the greater the number of stresses acting on the rice plant, the greater was the yield loss. Increasing pest abundance invariably led to greater maximum yield loss as all linear regressions had negative slopes, and the quadratic models eventually did at higher pest densities. As more stresses were applied, the result was more linear than quadratic regressions, and those regressions were less differentiated indicating lower N response and thus less compensation. Weeds and reduced solar radiation of the wet season were particularly suppressive. When these stresses were applied in the same treatment, yield

Nitrogen is the most required macronutrient for rice plants, particularly in double cropping systems. Due to the large requirement, concentrated inorganic sources, such as urea, are preferred by farmers. N results in two opposing interactions regarding pests. Numerous studies (Litsinger 1994) have shown that its application predisposes rice plants to insect injury and stimulates higher insect reproduction and feeding. Regarding stemborers N facilitates the entry of young larvae into the plant by stimulating tiller elongation that reduces the density of protective silica bodies (Bandong and Litsinger 2005). Leaffolders also increase in density in response to N application (Litsinger 1994). On the other hand, N deficiency reduces the capacity of the rice crop to tolerate insect pest damage, thus adding it greatly improves compensatory ability to overcome injury from the chronic pests included in this study. When the same infestation regimes were applied to each of the four rates of N, invariably the highest yields

114

J.A. Litsinger et al. average of a sample of fields within the recommendation domain. Previously, we showed that action thresholds differed for the three rice growth stages for stemborers and leaffolders (Litsinger et al. 2006b, 2006c). In the current study, damage functions likewise differed for reproductive and ripening stages (Tables 2 and 3). While peak attack is often confined to one growth stage in the field, infestation normally occurs in the other two stages, although it may not reach a level that would trigger an insecticide application. The companion study shows that sub-threshold densities, when combined with other insect pest injury, can lead to economic loss. We saw that when two growth stages were attacked, crop stress was exacerbated. When action thresholds were determined for stemborer and leaffolder, crops more often than not were also injured to some degree in both stages. The same was true with whorl maggot and defoliators where action thresholds were generated for each pest separately (Litsinger et al. 2006a). At times only one pest of the complex is present during in the vegetative stage. Only by using artificial infestation is it possible to isolate injury to specific growth stages. Having EILs for individual and dual stage infestations means that crop managers can be more efficient in their usage. However, as this study has emphasised, the aforementioned subtleties between pests and crop stages can be overwhelmed by other stresses acting on the crop which will significantly affect EIL values. We have examined only some of the most common stresses that can occur in rice fields. We will only be able to have a fuller understanding of the interactions of other stresses by expanding field studies or improving the reliability of crop modelling. On the other hand, experience has shown that in most locations only a few major stresses are important in any one season. If any of those stresses were those included in this study entomologists can utilise our damage functions directly or interpolate them to fit existing conditions. For example with regard to day-length and solar radiation based, the latitude of our study site was 158 27’ 11" N. If N dosages differ, they can likewise be adjusted based on our results. The EIL value is the slope taken from a linear regression. From quadratic equations, the linear portion is found subjectively by drawing a line with a pencil and ruler on the graph itself. Values can then be generated from the graph to produce a linear regression to calculate the slope. Another variable needed to derive EILs is the percentage of insect control expected for a corrective insecticide action. We have derived these values in the action threshold study (Litsinger et al. 2005). For whorl maggot one can expect 28% control from single spray or 50% if two sprays. Also from single sprays, low levels of control also occurred for the other pests: defoliators 50%, leaffolder 53%, and stemborer 40%. As costs and prices vary year to year we have not calculated EILs herein, but with the data we present, little effort would be needed to

level greatly declined. We noted the most and often steepest linear regressions were at the lowest yield levels for all three pest guilds in Figures 1F, 3F, and 5F. Even cursory examination of the figures shows that for each stress eliminated, the crop responded with higher yield level or less steep regressions. The least stress occurred in weeded, dry-season crops at the highest N level and lowest pest densities. It is difficult to conclude from this study as to which stress was most severe. The important point regarding management is which stresses can the farmer manage best? 4.5. Calculating the EIL

As we have seen, damage functions for common rice insect pests vary greatly due to many causes including the specific and number of crop growth stages attacked, other pests attacking the same growth stage, crop management practices, and the presence of other abiotic crop stresses. On the other hand, modern rice plants can often tolerate such attacks, particularly at low damage levels, if harmed only by a single insect pest, and when stresses such as N deficiency, weeds, and solar deficiency are less burdensome. Therefore as damage functions are highly condition specific, EILs are likewise highly condition specific (Poston et al. 1983). This is also borne out by the differing economic threshold values for the same pest in the literature (Way et al. 1991). It should be noted that our current study, replications comprised different farmers’ fields, separated by up to 1 km. Damage functions were therefore an average of six farms. This is relevant, as van Haltern (1979) calculated linear regressions for white stemborer (Scripophaga innotata [Walker]) whiteheads separately from 15 different fields sown to the same rice variety in Sulawesi, Indonesia. All of the regressions were linear, but each differed in slope where the percentage loss per 1% whitehead infestation (the damage function) ranged from 0.29 to 2.08, a difference of seven times between neighbouring fields. In Japan, Ishikura (1967) cited a study that covered 100 fields where 50 hills/field were purposely sampled so the hills covered the range of extant damage levels. The study was done in one season but included 11 varieties. Damage functions were generated per field which resulted in three different mathematical models, which when pooled gave a quadratic model. Therefore mathematical models of best fit can even deviate from farm to farm in the same season. Ishikura (1967) even found, as we documented in this study, damage functions varied by plant growth stage. In Japan this meant that they differed by pest generation, as stemborers emerge from dormancy during a brief period resulting in population synchrony. This is less common in the tropics where pest generations often overlap due to constant immigration, thus damage cannot be attributed to a specific pest generation. As pest control recommendations realistically cannot be made for individual fields it is prudent to take the

International Journal of Pest Management accomplish this. These data illustrate that insecticide control of chronic rice pests is a rather blunt instrument, thus it would be more prudent if the farmer instead spend his resources on crop management practices that strengthened the rice crop to tolerate insect pest injury. Acknowledgements
For both the papers in this series, we appreciate the generous cooperation provided by the farmers in the study sites. Many locally hired project staff were responsible for conducting the trials and their invaluable contributions are acknowledged. Those assisting in Zaragoza were Catalino Andrion and Rodolfo Gabriel and in Guimba George Romero. The assistance and advice from scientists at IRRI: Dr. Serge Savary, Francisco A. Elazegui, and Nestor G. Fabellar of the Plant Pathology Department and Ernesto G. Castillo in the Agronomy Department and Grace L. Reyes in the Statistics Department are gratefully appreciated.

115

References
Bandong JP, Litsinger JA. 2005. Rice crop stage susceptibility to the rice yellow stemborer Scirpophaga incertulas (Walker) (Lepidoptera: Pyralidae). Int J Pest Manage. 51:37–43. Capinera JE, Roltsch WJ. 1980. Response of wheat seedlings to actual and simulated migratory grasshopper defoliation. J Econ Entomol. 73:258–261. Dang TH. 1986. Damage and yield loss caused by stem borers in irrigated rice in Kenya. Trop Agric. (Trinidad) 63:201–204. DeDatta SK. 1981. Principles and practices of rice production. New York, NY: John Wiley and Sons. p. 618. Dyck VA, Than Htun, Dulay AC, Salinas GD, Jr, Orlido GC. 1981. Economic injury levels for rice insect pests. Agric Res J Kerala. 19:75–85. Gomez KA, Bernardo RC. 1974. Estimation of stem borer damage in rice fields. J Econ Entomol. 67:509–513. van Haltern P. 1979. The insect pest complex and related problems of lowland rice cultivation in South Sulawesi, Indonesia. Mededelingen Landbouwhogeschool, Wageningen, Netherlands. 79–1:1–111. Hull LA, Beers EH, Grimm JW. 1985. Action thresholds for arthropod pests of apple. In: Frisbie RE, Adkisson PL, editors. Consortium of IPM (CIPM) – Integrated Pest Management on Major Agricultural Systems. Texas Agricultural Experiment Station, Miscellaneous Publication No. 1616, College Station, Texas. p. 224–294. International Rice Research Institute (IRRI). 1983. Insect combinations. p. 202–204. In: Annual Report for 1982. Philippines: Los Banos. p. 532. ˜ IRRI. 1984. Yield losses caused by multiple species infestations. p. 187–188. In: Annual Report for 1983. Philippines: Los Banos. p. 494. ˜ Ishikura H. 1967. Assessment of rice loss caused by the rice stem borer. p. 251–264. In: The Major Insect Pests of the Rice Plant. Baltimore (MD): John Hopkins Press. p. 729. Jahn GC, Litsinger JA, Chen Y, Barrion AT. 2007. Integrated Pest Management of Rice: Ecological Concepts. p. 315–366. In: Koul O, Cuperus GW, editors. Ecologically based integrated pest management. Wallingford, UK: CAB International. p. 462. Kenmore PE, Carino FO, Perez CA, Dyck VA, Gutierrez ˜ AP. 1984. Population regulation of the rice brown ˚ planthopper (Nilaparvata lugens Stal) within rice fields in the Philippines. J Plant Protect Trop. 1:19–37.

Litsinger JA. 1993. A farming systems approach to insect pest management for upland and lowland rice farmers in tropical Asia. p. 45–101. In: Altieri MA, editor. Crop protection strategies for subsistence farmers. Westview Studies in Insect Biology. Boulder, CO: Westview Press. p. 197. Litsinger JA. 1994. Cultural, mechanical, and physical control of rice insects. In: Heinrichs EA, editor. Biology and management of rice insects. New Delhi: Wiley Eastern Ltd. p. 549–584. Litsinger JA. 2009. When is a rice insect a pest: yield loss and the Green Revolution. p. 391–498. In: Peshin R, Dhawan AK, editors. Integrated pest management: innovationdevelopment process. Berlin: Springer Science þ Media B.V. p. 670. Litsinger JA, Bandong JP, Canapi BL (2011). Effect of multiple infestations from insect pests and other stresses on irrigated rice in the Philippines: II. Damage and yield loss. Int J Pest Manage. Litsinger JA, Bandong JP, Canapi BL, dela Cruz CG, Pantua PC, Alviola III AL, Batayan E. 2005. Evaluation of action thresholds against chronic insect pests of rice in the Philippines: I. Less frequently occurring pests and overall assessment. Int J Pest Manage. 51:45–61. Litsinger JA, Canapi BL, Bandong JP, dela Cruz CG, Apostol RF, Pantua PC, Lumaban MD, Alviola III AL, Raymundo F, Libetario EM, Loevinsohn ME, Joshi RC. 1987. Rice crop loss from insect pests in wetland and dryland environments of Asia with emphasis on the Philippines. Insect Sci Appli. 8:677–692. Litsinger JA, Canapi BL, Bandong JP, dela Cruz CG, Apostol RF, Pantua PC, Lumaban MD, Alviola III AL, Raymundo F, Libetario EM, Loevinsohn ME, Joshi RC. 2006a. II. Whorl maggot and defoliators. Int J Pest Manage. 52:167–180. Litsinger JA, Canapi BL, Bandong JP, dela Cruz CG, Apostol RF, Pantua PC, Lumaban MD, Alviola III AL, Raymundo F, Libetario EM, Loevinsohn ME, Joshi RC. 2006b. III. Leaffolders. Int J Pest Manage. 52:181–194. Litsinger JA, Canapi BL, Bandong JP, dela Cruz CG, Apostol RF, Pantua PC, Lumaban MD, Alviola III AL, Raymundo F, Libetario EM, Loevinsohn ME, Joshi RC. 2006c. IV. Stemborers. Int J Pest Manage. 52:194–207. Litsinger JA, Libetario EM, Canapi BL. 2009. Eliciting farmer knowledge, attitudes, and practices in the development of integrated pest management programs for rice in Asia. p. 119–273. In: Peshin R, Dhawan AK, editors. Integrated pest management: dissemination and impact. Berlin: Springer Science þ Media B.V.. p. 634. Murugesan S, Chelliah S. 1983. Rice yield loss caused by leaffolder damage at tillering stage. Int Rice Res Newslett. 8(4):13–14. Nakasuji F, Dyck VA. 1984. Evaluation of the role of Microvelia douglasi atrolineata (Bergoth) Heteroptera: Veliidae) as a predator of the brown planthopper ˚ Nilaparvata lugens (Stal) (Homoptera: Delphacidae). Rese Popul Ecol. 26:134–140. Onstad PW. 1987. Calculation of economic-injury levels and economic thresholds for pest management. J Econ Entomol. 80:297–303. Ooi PAC, Shepard BM. 1994. Predators and parasitoids of rice insect pests. p. 585–611. In: Heinrichs EA, editor. Biology and management of rice insects. New Delhi: Wiley Eastern Ltd. p. 779. Pantoja A, Smith CM, Robinson JF. 1986. Effects of the fall armyworm (Lepidoptera: Noctuidae) on rice yields. J Econ Entomol. 79:1324–1329. Pedigo LP, Hutchins SH, Higley LG. 1986. Economic injury levels in theory and practice. Annu Rev Entomol. 31:341– 368.

116

J.A. Litsinger et al.
Shiraki T. 1917. Paddy borer (Schoenobius incertellus Walker). Taihoku Agricultural Experiment Station, Formosa. p. 256. Soejitno J. 1977. Relation between damage by rice stem borer Tryporyza incertulas and yield of rice variety Pelita I-1. Int Rice Res Newslett. 2(4):6. Suwongwan P, Catling HD. 1987. Effects of high temperature and low humidity on the survival of eggs and larvae of yellow stem borer, Scirpophaga incertulas (Walker), in deepwater rice. J Plant Protect Trop. 4: 115–119. Turnipseed SG, Kogan M. 1987. Integrated control of insect pests. p. 779–817. In: Wilcox JR, editor. Soybeans: improvement, production and uses. American Society of Agronomy, Madison WI. p. 888. Viajante VD, Heinrichs EA. 1986. Rice growth and yield as affected by the whorl maggot Hydrellia philippina Ferino (Diptera: Ephydridae). Crop Protect. 5:176– 181. Viajante VD, Heinrichs EA. 1987. Rice grain yield as affected by the feeding of the green hairy caterpillar, Rivula atimeta (Swinhoe) (Lepidoptera: Noctuidae). J Plant Protect Trop. 4:27–35. Way MO, Grigarick AA, Litsinger JA, Palis F, Pingali PL. 1991. Economic thresholds and injury levels for insect pests of rice. p. 67–105. In: Heinrichs EA, Miller TA, editors. Rice insects: management strategies. Berlin: Springer Science. Yoshida S. 1981. Fundamentals of rice crop science. Los Banos, Philippines: IRRI. p. 269. ˜

Poston FL, Pedigo LP, Welch SM. 1983. Economic injury levels: reality and practicality. Bull Entomol Soc Am. 29:49–53. Reissig WH, Heinrichs EA, Litsinger JA, Moody K, Fiedler L, Mew TW, Barrion AT. 1986. Illustrated guide to integrated pest management in rice in tropical Asia. IRRI, Los Banos, Philippines. p. 411. ˜ Rice SE, Grigarick AA, Way MO. 1982. Effect of leaf and panicle feeding by armyworm (Lepidoptera: Noctuidae) larvae on rice grain yield. J Econ Entomol. 75:593–595. Rombach MC, Roberts DW, Aguda RM. 1994. Pathogens of rice insects. In: Heinrichs EA, editor. Biology and management of rice insects. New Delhi: Wiley Eastern Ltd. p. 613– 655. Rothschild GHL. 1971. The biology and ecology of rice-stem borers in Sarawak (Malaysian Borneo). J Appl Ecol. 8:287–322. Savary S, Elazequi FA, Moody K, Litsinger JA, Teng PS. 1994. Characterization of rice cropping practices and multiple pest systems in the Philippines. Agric Sys. 46:385–408. Savary S, Srivastava RK, Singh HM, Elazegui FA. 1997. A characterisation of rice pests and quantification of yield losses in the rice-wheat system of India. Crop Protect. 16:387–398. Seth GR, Sardana MG, Khosla RK, Kalyanaraman VM. 1969. Preharvest losses due to pests and diseases in rice (Oryza sativa L.). Ind J Agric Sci. 39:1113–1124. Shepard BM, Justo HD, Jr, Rubia EG, Estano DB. 1990. ˜ Response of the rice plant to damage by the rice whorl maggot Hydrellia philippina Ferino (Diptera: Ephydridae). J Plant Protect Trop. 7:173–177.