Upload
james-litsinger
View
219
Download
0
Embed Size (px)
Citation preview
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
1/17
Evaluation of action thresholds for chronic rice insect pests in the
Philippines. I. Less frequently occurring pests and overall assessment
J. A. LITSINGER1, J. P. BANDONG2, B. L. CANAPI3, C. G. DELA CRUZ2, P. C. PANTUA2,
A. L. ALVIOLA2, & E. H. BATAY-AN III4
11365 Jacobs Place, Dixon, CA, USA,
2International Rice Research Institute, Metro Manila, Philippines,
3Monsanto
Philippines, Makati, Metro Manila, Philippines, and4
Philippine Department of Agriculture
Abstract
Action thresholds as decision tools for insecticide application were developed and tested against the major insect pests of rice
at four sites in the Philippines over a 13-year period. Action threshold treatments were compared to the farmers practice,prophylactic insecticide usage, and an untreated check. Yield loss data using the insecticide check method partitioned yieldlosses over three crop growth stages in the same test fields. Chronic pests that exceeded action thresholds in 79% of fieldswere whorl maggot Hydrellia philippina Ferino (Diptera: Ephydridae), defoliators Naranga aenescens Moore and Rivulaatimeta (Swinhoe) (Lepidoptera: Noctuidae), leaffolders Cnaphalocrocis medinalis (Guenee) and Marasmia patnalis Bradley(Lepidoptera: Pyralidae), and stemborers Scirpophaga incertulas (Walker) and S. innotata (Walker) (Lepidoptera: Pyralidae).Minor chronic pests reached threshold levels in only one site each: rice bug Leptocorisa oratorius (F.) (Koronadal),whitebacked planthopper Sogatella furcifera (Horvath) (Zaragoza) and green leafhopper Nephotettix virescens (Distant)(Guimba); brown planthopper Nilaparvata lugens (Stal) did not exceed a threshold in any field. Stemborers were the mostimportant pest group in terms of yield loss. Despite the insecticide check method underestimating losses, a mean crop loss of0.62 t/ha was measured which showed ample scope for corrective action. But loss was evenly distributed across crop growthstages (0.15 0.24 t/ha) reducing the impact of insecticides. Action threshold treatments overall outyielded the untreatedcheck, more so in the two sites with highest pest density. The benefit of thresholds was to reduce insecticide usage, as a costsaving. However all the practices showed poor economic returns including the farmers practice. Farmers practice employedlow insecticide dosages and timing was not consistent with pest damage, but yields were often similar to threshold treatments.
Farmers appear to use insecticide more for risk aversion than for profit. The best threshold characters when evaluated againstresulting pest density and yield loss criteria showed accuracies 490% correct decisions. Future work is needed to improvethe insecticide response rather than monitoring tools. Thresholds need to be incorporated into improved crop management,which was often found suboptimal by farmers, to take advantage of the high levels of tolerance in modern high tilleringcultivars. Crop husbandry practices which improve yield potential such as selection of longer maturing varieties and nitrogenfertilizer may be a more effective pest management strategy than insecticides.
Keywords: Pest control, irrigated rice, insecticides, decision making, yield loss, plant tolerance, planting date, colonisation
pattern, nitrogen fertilisation
1. Introduction
Modern rice varieties possessing genetic resistance
against brown planthopper Nilaparvata lugens (Stal)
and green leafhopper Nephotettix virescens (Distant),
the major epidemic insect pests, have been widely
grown by farmers since the mid 1960s (IRRI 1985).
Genetic resistance against chronic insect pests has
been less successful (Heinrichs 1986). Whorl maggot
Hydrellia philippina Ferino, defoliators Naranga
aenescens Moore and Rivula atimeta (Swinhoe),
leaffolders Cnaphalocrocis medinalis Guenee and
Marasmia patnalis (Bradley), stemborers Scirpophaga
incertulas (Walker) and S. innotata (Walker), rice bug
Leptocorisa oratorius (F.), and whitebacked planthop-
per Sogatella furcifera (Horvath) are recurring pestscausing significant yield losses in the Philippines as
determined by the insecticide check method (Lit-
singer et al. 1987). Shortages in rice production in
the 1970s and 1980s due to epidemic pests spurred
the development of integrated pest management
(IPM) strategies with the focus on minimising
insecticide usage, which had been found to upset
natural enemy populations leading to pest resurgence
and secondary pest outbreaks. It is ironical that
farmers now often overuse insecticide from fear of
losses caused by a past history of crop failures from
epidemic pests (Kenmore et al. 1987; Litsinger
1989). Farmers also overestimate losses from chronic
pests based on their experience from epidemic pests
prompting further overuse (Heong and Escalada
1997).
Chronic insect pests are the main targets for IPMtraining programmes now widespread in Asia where
Correspondence: J. A. Litsinger, 1365 Jacobs Place, Dixon, CA 95620, USA. Tel: 1 707678 9068. Fax: 1 707678 9069. E-mail: [email protected]
International Journal of Pest Management, January March 2005; 51(1): 45 61
ISSN 0967-0874 print/ISSN 1366-5863 online # 2005 Taylor & Francis Group Ltd
DOI: 10.1080/09670870400028284
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
2/17
farmers learn crop monitoring and management
decision skills (Matteson 2000). One of the basic
tenets of IPM is to minimize insecticide usage, thus
pest populations are tolerated until pest density has
reached an economic threshold level or when other
control methods are impractical or uneconomical(Norton and Mumford 1993). The main tools for
insecticide decisions are economic thresholds which
are pest densities that trigger a corrective action
before the damage reaches the critical economic
injury level (Morse and Buhler 1997). Economic
thresholds are based on damage relationships be-
tween abundance and yield which along with
economic parameters can predict the most optimal
new thresholds as costs change. Damage functions
have not been worked out for chronic rice insect
pests, thus empirically derived action thresholds
(ATs) have been utilised along with surveillance
methods (Dyck et al. 1981; Reissig et al. 1986; Wayet al. 1991).
In the developed world, farmers often contract
surveillance activities to professional scouts to make
pest control decisions. In the developing world,
small-scale farmers cannot afford to hire scouts, thus
the use of ATs must be able to be mastered by
farmers. Thresholds have often been said to be too
difficult for farmers to master (Goodell 1984;
Matteson et al. 1994; Morse and Buhler 1997), as
those developed by researchers usually are not in a
form that farmers understand. Goodell et al. (1982)
advocated that for IPM technology to be easilyassimilated, research should start with the farmers
and not from research stations. Studies have shown
that most farmers develop their own thresholds and
monitoring techniques (Bandong et al. 2002). A
number of the threshold characters being evaluated
in this study were ideas that came from farmers
(Bandong et al. 2002). A reciprocal relationship
between the farmers and researchers perception of
ATs needs to be fostered as both stand much to gain
from each others perspectives and individual skills
(Goodell 1984; Heong and Escalada 1997).
This study represents the field testing of ATs
against chronic insect pests of rice following on from
the early work reported in Heinrichs et al. (1978),
Waibel (1986), and Smith et al. (1988). ATs are
composed of a character to measure, a level for that
character, a monitoring protocol, and a corrective
response, normally an insecticide that would entail a
recommended dosage and timing. ATs were com-
pared to the farmers practice, a prophylactic best
insecticide practice, and an untreated check.
The resilience of rice crops to pest damage was
evaluated by site and season using yield loss data.
There are reports of exceptional examples of modern
rices ability to compensate from abnormally highinsect infestation: crops suffering 50 and 82% rice
whorl maggot damaged leaves (Viajante and Hein-
richs 1986; Shepard et al. 1990), 67% damaged
leaves from leaffolder (Miyashita 1985), 30% dead-
hearts and 10% whiteheads (Rubia et al. 1996), and
three whiteheads/hill (Litsinger 1993). Nitrogen (N)
contributes to this resilience and was tested as a
contrasting strategy to insecticide in response to ATs.
This paper is the first of a four-part series. The
succeeding papers focus on the development for ATsfor whorl maggot and defoliators (Litsinger et al.
2005) while those on leaffolders and stemborers will
follow. This first part reports on the ATs for the less
common pests and gives an overview.
2. Materials and methods
2.1. Study sites
ATs were tested on-farm over a 13-year period in
four irrigated, double-crop rice areas typical of the
Philippines major rice bowls (Pingali et al. 1997)
23 crops in Zaragoza (141 fields), 15 crops inKoronadal (109 fields), 13 crops in Guimba (88
fields), and 17 crops in Calauan (81 fields). Farm
sizes ranged from 51 to 4 ha with land preparation
done by rotary tillers. Two sites Zaragoza (mean
2.1-ha farms) and Guimba (mean 0.9-ha farms)
were in Nueva Ecija province, Central Luzon, while
Calauan (mean 2.1-ha farms) lies in Southern
Luzon, Laguna province. A fourth site, Koronadal
(mean 1.0-ha farms), South Cotabato province is in
Mindanao southern Philippines.
Zaragoza comprised the villages of Marawa,
Malabon Kaingin, Imbunia, Rajal Norte, and Bati-tang in portions of Zaragoza, Jaen, and Santa Rosa
towns. Zaragoza farmers fields lie at the lower end of
the Upper Pampanga River Irrigation System and
consequently were planted late in the wet season due
to delayed water delivery. Crops often reach maturity
during the time of the strongest typhoons (between
October 15 and November 15) and as a result suffer
severe damage from flooding or lodging, or indirectly
by diseases spreading from wind-whipped foliage or
due to harvested grain that cannot be dried. Further
description of farmers and the area is given by
Goodell et al. (1982).
The Guimba site was located in the village of
Bantug next to the International Rice Research
Institute (IRRI) and Philippine Regional Department
of Agriculture Cropping Systems Program research
site in nearby San Roque and Macatcatwit. Deep
well pumps, one per village, each irrigated 60 100
ha. Due to the small irrigation systems farmers plant
their wet season crop earlier than in Zaragoza,
avoiding the severe typhoons. Tungro disease is
endemic, occurring sporadically. The electric pumps
are expensive to operate and break down often
leading to drought stress. Further site description is
given in Entomology Department (1985).Calauan is located 10 km south of the IRRI
research centre in the villages of Pulong and Dayap.
The rice area is irrigated by river diversion from small
streams feeding Laguna de Bay lake from the
46 J. A. Litsinger et al.
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
3/17
watershed of Mt. Makiling volcano. The villages are
located along the lake edge but away from the
perennial flood zone. Calauan farmers experienced
large scale tungro outbreaks in the early 1970s. In
Koronadal the villages of Namnama, Magsaysay,
Barrio 1, Avancenia, and Morales were selected inthe Marbel River Irrigation System. Koronadal lies
outside of the typhoon belt but has been the site of
outbreaks of tungro, grassy stunt, and brown
planthopper hopperburn in the early 1980s. With
its longer rainy season, the double crops are referred
to as first and second rather than wet season and
dry season. Its volcanic soil is still fertile and high
yields are obtained with lower amounts of inorganic
fertilizer. Further site description is given in En-
tomology Department (1985, 1988).
2.2. Research teamsATs were developed and tested in farm commu-
nities by resident research teams for each site. Field
workers were recruited from the surrounding villages
representing major ethnic groups. A field office was
rented and staffed by a laboratory assistant trained in
ricefield arthropod identification and equipped with a
dissecting microscope. The resident entomologist
and team were encouraged to improve on current
thresholds with farmers as co-partners. Thus there
was constant revision of threshold characters, levels,
insecticides, and methods of application.
2.3. Experimental design
Thresholds were tested under farmers agronomic
conditions as much as possible including their
currently used cultivars and seedbed methods. The
only stipulation made was not to include direct
seeded crops as this has been found to affect pests
(e.g., whorl maggot Litsinger 1994). The purpose
was to conduct the trials under farmers field
conditions and to only vary the insect control
variable. The many cultivars used, for example,
possessed genetic resistance against a similar com-
plex of insect pests and diseases. The dapog seedbed
method adds 2 weeks to the crop in the field.
Farmers vary their management practice as they see
fit throughout the season more commonly than
following a standard set of practices. We did not
attempt to dictate crop management practices for
them. The results of the research will yield technol-
ogy that is more robust and adaptive.
Each season four to nine new farmers (replica-
tions) were identified from each site to be co-
operators. These farmers were dispersed geographi-
cally within each community and grew the most
popular variety at the time. Planting dates werestaggered over the planting season to allow expres-
sion of the full range of pest densities. The most
popular varieties changed over time but included
IR36, IR42, IR62, IR64, IR74 (maturities ranging
from 90 to 110 days) in most sites. Locally farmers
had other selections: Koronadal with IR60 (released
only in Mindanao) and an unofficial line #90;
Calauan with C1 and Malagkit (both 120 days) and
IR70; Guimba with IR58; Zaragoza with IR52 and
IR56. To be released, Philippine varieties must beresistant to green leafhopper and brown planthopper
which caused severe epidemics in the 1970s before
improved multiple pest-resistant varieties were devel-
oped. As resistance is not durable, new varieties must
be developed continually.
Crops were transplanted from wet seedbeds or
dapog. The latter was popular in Calauan and
Koronadal where seeds are sown on banana leaves
thus roots do not enter soil to minimize transplanting
shock (from roots torn off while pulling seedlings),
field establishment occurs 2 weeks earlier than a
wetbed. Each season 30 40 farmers, including the
current co-operators, were interviewed every fewweeks to record their cultural practices (e.g.,
Department of Entomology 1985, 1988).
Within each co-operators field, a 0.2-ha research
area was demarcated with plastic string tied to
bamboo stakes (Litsinger et al. 1980a). Typically
eight to 10 treatments were included in the experi-
mental design each season combining yield loss
assessment along with two thresholds compared to
the farmers practice, a prophylactic insecticide
regime, and an untreated check. With the exception
of the farmers practice (1000 m2
), plot sizes were
100200 m2
. This design and plot size arrangementhave been determined to have unbiased effects on
pest abundance (Litsinger et al. 1987). Five of the
treatments were designed to measure yield loss using
the insecticide check method which had been tested
earlier in other sites (Litsinger 1991). Three crop
stages were recognised following Yoshida (1981)
vegetative (transplanting to panicle initiation), repro-
ductive (panicle initiation to flowering), and ripening
(flowering to maturity 10 days before harvest). In a
typical 110-day variety, the reproductive stage would
begin about 40 d.a.t. and end about 30 days later.
The seedbed had been included in experiments prior
to the current study, but as no significant yield loss
was ever measured, the seedbed portion of the
vegetative growth stage was eliminated.
The first treatment termed full protection at-
tempted to show the yield potential with the least
possible insect damage. Weekly insecticide sprays at
the manufacturers recommended dosages were
applied to each plot with interplot spray drift
minimised by a mosquito cloth on a 1 6 3-m wood
frame held downwind by two assistants. Insecticides
were selected both for their efficacy as well as proven
neutrality regarding phytotoxic or phytotonic effects
on rice (Venugopal and Litsinger 1984). Insecticideswere applied as foliar sprays with 19-l, lever-operated,
knapsack sprayers fitted with hollow cone nozzles
using a 200 300-l/ha spray volume (increased with
crop growth). From the second to fourth treatments,
Action thresholds for chronic rice insect pests 47
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
4/17
insecticide was withheld from each successive growth
stage, with the fifth treatment being the untreated
check. The prophylactic treatment involved soil
incorporation of 0.5 kg a.i. carbofuran granules/ha
before transplanting followed by two foliar sprays of
0.4 kg a.i. chlorpyrifos/ha 10 days apart during thelate reproductive stage to prevent stemborer white-
heads. All treatments were in randomised complete
block design with replications as fields. Yield was
taken from five 5-m2
crop cuts in a stratified sample
per treatment and dried to 14% moisture.
Total yield loss was calculated both in terms of grain
weight and percentage. Total yield loss was the
difference between the full protection and untreated.
To calculate percentage, total loss was divided by the
untreated yield and multiplied by 100. Loss in each of
the three growth stages was calculated in separate
treatments where protection was omitted from each
stage sequentially. Yield was subtracted from that ofthe full protection treatment. The loss in each of the
growth stages was summed and adjusted upwards or
downwards proportionally so that the total of the three
stages equalled the total yield loss. Cropcompensation
was measured by regression analysis using the yield
loss dataset for each of the two seasons per location
using crop averages that varied over a range of yield
potentials. Compensation would occur if the rate of
yield loss did not rise proportionally with increasing
yield (i.e., slope of the linear regression equation was
insignificant).
2.4. Sampling methods
Pest incidence was sampled weekly in the thresh-
old treatments and untreated check, but only once
per growth stage in the yield loss treatments. Pest
monitoring for AT decision making was carried out
in the respective threshold plots. Pest or damage
densities were usually measured on a per-hill basis
with the sample size of 20 hills selected individually
in a stratified pattern exclusive of a 1-m border zone.
Mechanical hand counters were used to record the
number of tillers and leaves per hill with pest damage
recorded on those plant parts as appropriate. One
staff scored the hill while another recorded.
2.5. Action thresholds
Thresholds are multifaceted involving a number of
variables, any one of which can affect efficacy. The
first variable is a character such as an insect stage
(adult, nymph) or its damage symptom (damaged
leaves, deadhearts). Second is the sampling unit and
number of samples that express the character
(planthoppers = 20 hills, green leafhopper = 25 net
sweeps, rice bug = per hill). Third is the density of thecharacter per sampling unit (e.g., one planthopper or
rice bug per hill, one leafhopper per sweep).
Normally a single character with two threshold levels
was tested each season per site, termed low level
(e.g., one planthopper per hill) and high level (e.g.,
two planthoppers per hill), in separate treatments.
New characters were continually being developed in
an effort to improve performance.
Another set of variables is associated with the
corrective response triggered by a threshold, usuallyan insecticide or N. As development of ATs was
iterative there was no balanced design to test the
many characters and response variables in a given
field. Most characters were tested in multiple sites.
Data analysis after each season entailed comparing
yield in the threshold treatments to that in the
untreated check, farmers practice, and prophylactic
treatment. An economic analysis was performed on
each practice where marginal returns and benefit:cost
ratios were calculated. Included in the analyses were
crop monitoring and insecticide application labour,
cost of labour, and interest. The farmers practice
was what each individual farmer collaborator carriedout in the trial field. Results were scrutinised to
determine if yield loss occurred in each growth stage
where thresholds were reached. If no yield loss were
recorded but thresholds were reached, levels were
raised the following season and vice versa. Figures
were drawn to illustrate the weekly pest abundance
and degree of control obtained field by field (e.g.,
Department of Entomology 1985, 1988).
2.5.1. Planthopper thresholds. The whitebacked and
brown planthoppers were monitored weekly in the
vegetative and reproductive stages by recording thenumber of adults and nymphs as well as their predators
from 20 hills along a zigzag transect in each plot
(Matteson et al. 1994). Each hill was bent over the
water with one hand and slapped with the other open
hand to dislodge arthropods including predators onto
the water surface for easier detection. The average
number of tillers per hill was calculated. Only the
brownish mature nymphs (stadia 4 5) and adults
were counted. Densities of immature white nymphs
are particularly prone to predation. Predator density
was incorporated into the ATs, as for each found, five
planthoppers were subtracted from the total. The AT
levels ranged from 0.5 to 1 planthoppers (adults and
nymphs) per tiller. Once half of the AT level were
reached, monitoring was increased to twice a week.
Fenobucarb (BPMC) insecticide was applied at 0.4 kg
a.i./ha timed to when the planthopper population was
in the late nymphal stage. Timing with mature nymphs
was to minimize the egg load inside the tillers, as the
insecticide has only minimal activity against eggs. If
the crop were to be sprayed when adults were
dominant, adults would be killed but eggs would
survive inside the tillers to emerge under very low
predator densities, often leading to resurgence.
2.5.2. Green leafhopper threshold. Green leafhopper
was only monitored if tungro virus was observed in
the community. The orange yellow discolouration,
diagnostic of tungro, is masked in mature plants but
48 J. A. Litsinger et al.
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
5/17
becomes particularly noticeable in regrowth after
harvest. A sweepnet was used in the seedbed through
the vegetative stage, taking 25 strokes while walking.
The AT level was set at 0.5 adults + nymphs/sweep
with 0.4 kg a.i. fenobucarb/ha as the response.
2.5.3. Ricebug threshold. Twenty hills were inspected
during the milk stage of grain development for rice
bug adults and nymphs. The range of AT levels tested
was four to 10 bugs/20 hills with a response of 0.4 kg
a.i. endosulfan/ha as a single spray.
2.6. Insecticide response
When a threshold were reached and insecticide
was the given response, it was applied within a day of
monitoring. Insecticide technology was developed in
an iterative process as well, thus if efficacy was low,
adjustments were made, normally changing thechemical but in certain instances involved research
on other application methods. Tests to determine the
minimum effective dosage were undertaken with a
view to cost savings. Foliar sprays were applied as
described in the yield loss trials. Percentage control
of each threshold character was based on the
untreated check.
2.7. Nitrogen substitution
Application of N has been shown to increase the
rice crops tolerance of pest damage (Litsinger 1993,1994; Rubia et al. 1996) and was tested in one of the
threshold treatments in all four sites during the final
2 years as a substitute for insecticide. The motive was
derived from the often discouraging results with
insecticides. Urea was broadcast at 25 kg N/ha,
equivalent in cost to an average insecticide applica-
tion, in response to surpassing an AT for whorl
maggot (eggs per hill from a neighbouring field),
defoliators (larvae per hill), leaffolders (larvae per
hill), and stemborers (egg masses per m2
).
2.8. Statistical analysis
All statistical analyses were performed by SAS and,
unless otherwise stated, we use P 4 0.05 as the
criterion for significance. Results were subjected to
ANOVA and regression/ correlation analysis where
appropriate. Treatment means were separated using
the paired t-test for two variables or least significant
difference (LSD) test for more than two variables.
Means are shown with standard errors of the mean
(SEM) using a pooled estimate of error variance.
3. Results
3.1. Insect pests and densities
The chronic pests that regularly exceeded thresh-
olds were whorl maggot, defoliators, leaffolders, and
stemborers (Scirpophaga incertulas was prevalent in
all sites except Koronadal where S. innotata was
dominant). Less common pests that reached thresh-
old levels only in single sites were rice bug in three
crops in Koronadal (5.8% of all test fields in the
site), whitebacked planthopper in two crops inZaragoza causing patches of hopperburn (9.1% of
fields), and green leafhopper in one crop in Guimba
with tungro symptoms occurring sporadically (0.8%
of fields). The latter three pests occurred too
infrequently for different threshold characters to be
tested. Brown planthopper affected 510 fields
nearby test fields causing isolated patches of
hopperburn on susceptible cultivars in all sites
except Calauan, but never in the test fields
themselves.
3.2. Yield lossMeasurement of yield loss is an essential element
in evaluating thresholds. With use of the insecticide
check method it was expected that the full protec-
tion treatment would provide 480% control for
each pest based on damage. This goal was only
achieved with leaffolders averaging 82.5+ 4.2%
damaged leaves. Control of damage by defoliators
(71.3+5.0% damaged leaves) and stemborers
(67.0+3.2% control based on deadhearts and
whiteheads) nearly reached the goal but the greatest
disappointment came with whorl maggot. Despite
weekly applications of high dosage foliar sprays, only55.2+5.3% control was achieved based on damaged
leaves over the four sites. Stemborers were the only
pest to show significant differences by season
(62.8+6.4 vs. 71.2+5.6% in the wet and dry
seasons, respectively, by paired t-test with P= 0.04,
df = 126).
Yield loss, despite the suboptimal insecticide
protection, showed considerable scope for IPM when
viewed as a total (0.62 t/ha or 12.7%) (Table I).
Yields across sites and seasons in the full protection
treatment averaged 4.99 t/ha. Highest seasonal yields
were recorded in Zaragoza dry season (6.23 t/ha) and
lowest in the Guimba wet season (4.39 t/ha), the two
closest sites. Untreated crops averaged 4.37 t/ha
overall with the highest and lowest site yields
occurring in the Zaragoza dry season (5.50 t/ha)
and in the Guimba wet season (3.67 t/ha). Lowest
yield per field was in Guimba 0.77 t/ha in the 1984
wet season. Highest yield loss per crop was also in
Guimba dry season (0.77 t/ha); lowest was in
Calauan wet season (0.30 t/ha).
Losses during the vegetative stage (0.23 t/ha) were
significant in all sites and seasons except the dry
seasons in Guimba and Calauan. No one site or crop
had significantly higher losses than another duringthis stage. The reproductive stage loss (0.24 t/ha loss)
was significant in all site season combinations
except Koronadal second crop and Calauan wet
season. Least loss occurred in Calauan wet season
Action thresholds for chronic rice insect pests 49
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
6/17
Table I. Measurement of yield loss by season in four irrigated rice sites by the insecticide check method, Philippin
Yield (t/ha) Yield lossa/
Crops Fields Full
Total Vegetative stage Reproductive s
Site Season3 (no.) (no.) protection Untreated t/ha % P t/ha % P t/ha %
Zaragoza WS 12 72 5.09+0.28 b 4.42+0.18 b 0.70+0.19 ab 12.8+2.8 50.0001 0.24+0.11 4.5+1.8 0.03 0.27+0.07 a 5.3+1.DS 11 69 6.23+0.21 a 5.50+0.20 a 0.63+0.11 bc 10.2+1.6 50.0001 0.25+0.06 4.0+0.9 0.005 0.23+0.06 ab 3.8+0.
Koronadal 1st 7 52 5.16+0.25 b 4.55+0.20 b 0.60+0.11 bc 11.3+1.8 50.0001 0.24+0.07 4.4+1.2 0.004 0.21+0.08 ab 4.1+1.2nd 8 57 4.85+0.19 b 4.10+0.15 b 0.75+0.11 ab 15.3+1.9 50.0001 0.32+0.05 6.8+1.3 0.0004 0.26+0.07 a 4.1+1.
Guimba WS 7 44 4.39+0.58 b 3.67+0.58 b 0.72+0.09 ab 21.9+7.9 50.0001 0.21+0.06 8.6+5.0 0.001 0.29+0.03 a 7.4+1.DS
6 44 4.80+
0.53 b 4.03+
0.53 b 0.77+
0.16 a 18.1+
4.85
0.0001 0.20+
0.06 5.0+
1.7ns
0.34+
0.07 a 8.1+
0Calauan WS 9 44 4.61+0.22 b 4.27+0.25 b 0.30+0.10 c 6.3+2.1 50.0001 0.18+0.08 3.9+1.7 0.03 0.06+0.02 b 1.4+0.DS 8 37 4.79+0.24 b 4.38+0.25 b 0.39+0.08 bc 8.4+1.8 50.0001 0.13+0.06 2.4+1.2 ns 0.16+0.06 ab 3.7+1.
total 68 419
avg 4.99+0. 12 4 .3 7+0. 12 0 .6 2+0.08 12.7+0.4 50.0001 0.23+0.07 4.8+0.3 50.0001 0.24+0.07 4.9+0.P 0.003 0.003 0.05 ns 0.02
F 3.5 3.64 2.13 0.49 2.62
df 67 67 67 67 67
1In a column, means+SEM followed by a common letter are not significantly different (P40.05) by LSD test. 2Probabilities show significant differences of losse
protection in that stage compared to the insecticide protected treatment. Total yield loss was measured as the difference between the untreated and insecticid
vegetative, reproductive, and ripening stages. 3WS= wet season, DS = dry season.
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
7/17
0.06 t/ha. The ripening stage loss (0.15 t/ha) was
significant in only half the site-crop combinations.
No one site or crop combination was significantly
different.
A wide range of yield losses and yield potentials
occurred within each site and season. With yield lossas the dependent variable, regressions were made for
each site and season (Figure 1). In Zaragoza a
distinct seasonal difference in compensation ability
was seen (Figure 1a). The wet season data showed an
increasing rate of yield loss with rising yield levels
leading to a significant regression showing a lack of
compensation. But in the dry season a high degree of
compensation was evident as crops with rising yield
potentials from 1.5 to 7 t/ha showed an insignificant
rise in yield loss.
Data from Koronadal (Figure 1b), however,
showed a lack of compensation in either the first
or second crops as the regressions were bothsignificant. Whereas in both wet and dry season
crops in the low pest density sites of Guimba and
Calauan (Figure 1c d) high rates of crop compen-
sation were evident.
3.3. Nitrogen substitution
The results of substituting insecticide with fertiliser
as a response to thresholds showed significant yield
gains over the untreated plots in three of the eight
pest season combinations (Table II). Due to the fewseasons of testing, the results were pooled over
locations. Only responses to stemborers showed a
yield gain in the wet season, while both whorl maggot
and defoliators did in the dry season. There were no
significant yield gains with leaffolder thresholds in
any season.
3.4. Threshold treatments compared to other practices
Thresholds, both high and low levels, were com-
pared to three other pest control strategies: (1) doing
nothing (the untreated check), (2) farmers practice
(among the farmer co-operators), and (3) prophylacticinsecticide scheduling. The full protection treatment
served as a check showing yield potential with
maximum insecticide protection. The variables ana-
lysed for each treatment were percentage of fields
Table II. Yield gain from pooled fields across sites where nitrogen fertiliser replaced the insecticide response to action thresholds 1.
Wet season Dry season
Pest Yield gain (kg/ha)2 P df Yield gain (kg/ha)2 P df
Whorl maggot 52+42 ns 23 389+43 0.001 24
Defoliators 121+57 ns 15 274+87 0.02 18Leaffolders 220+132 ns 25 250+112 ns 24
Stemborers 261+117 0.03 30 240+186 ns 19
1Nitrogen was applied as urea at 25 kg N/ha when a threshold was reached. 2Difference between threshold treatment and untreated by paired
t-test at P40.05, mean+SEM.
Figure 1a d. Relationship between yield and insect pest-caused yield loss in four sites showing compensatory capacity in those crops with
significant slopes in the regression equation: (a) Zaragoza, (b) Koronadal, (c) Guimba, and (d) Calauan, Philippines.
Action thresholds for chronic rice insect pests 51
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
8/17
treated with insecticide, application frequency of
treated fields, dosage load per crop, efficacy against
target pest groups, yield gain, total yield, marginal
returns from insect control, and benefit cost ratio.
Over all sites, ca. twice the percentage of fields were
treated with the low value than the high valuethresholds (68 vs. 36%) (Table III). This was much
less than the average for farmers (90%) and the
standardised prophylactic regime (100%). As deter-
mined from interviews, farmers in three of the sites
treated 492% of fields for an average crop, with
Guimba farmers treating the fewest (76%). Guimba
farmers may have tailored applications to the rela-
tively low pest densities recorded there, but, more
likely, they were more strapped for cash to purchase
insecticide than farmers in other sites due to the high
cost of running the electrical pump. Highest treat-
ment frequency occurred with Calauan farmers
(4 98%), also a low pest density site. Significantdifferences occurred within and between each thresh-
old treatment with regard to pest density. The
frequency of fields treated in the high pest density
sites with the low threshold level was equivalent to
farmers (90% or more) but was557% from the high
threshold treatment. But in the low pest density sites,
the low threshold resulted in significantly fewer (39
53%) fields being treated, and even fewer (525%)
with the high threshold level.
The same general trends were evident in terms of
the mean number of applications per field, but the
differences were less distinct. This statistic was,however, based on treated fields and not all fields.
Farmers averaged one more application (2.3 times)
than the high threshold treatment (1.3 times), with
the low threshold treatment closer to the high
threshold (1.6 times). This indicated that the thresh-
old levels may have been set too close together, and
thus were more likely to be exceeded in both
treatments in each field. Zaragoza was unusual as
there was no difference between the farmers practice
and the threshold treatments, all ranging between 1.9
and 2.0 times. This was due both to the greater whorl
maggot pressure and the response of double spray-
ings, as well as more than one pest exceeding
thresholds per field.
Greatest frequency occurred with Koronadal farm-
ers (3.1 applications per crop), probably due to
farmers responding to recent history of pest out-
breaks (Waibel 1986). In Guimba where only 76% of
farmers had treated, the number of applications was
equal to all sites but Koronadal. This indicated that
of the Guimba farmers who sprayed, they applied
more frequently than farmers in other sites. The
threshold treatments resulted in lower application
frequencies than the farmers practice in three sites
for the high thresholds and two sites in the lowthresholds. There was no statistical difference be-
tween threshold treatments in any site.
The most popular insecticides of farmers, account-
ing for 74% of total applications, were mono-
TableIII.Comparisonofcorrectiveactionstakenbetweenth
resholdsandfarmerspracticeinsecticideapplicationfrequency
1.
Fieldstreated(%)
Insectic
ideapplications(no./field)3
P
est
CropsFields
Low
High
Farmers
Low
High
Farmers
Site
de
nsity
(no.)
(no.)
threshold2
threshold2
practice
P
F
df
threshold
threshold
practice
P
F
df
Zaragoza
H
igh
22
126
89.5+6.9Aa
50.8+7.8Ab
92.5+3.2Aa
0.003
4.06
65
2.0+0.4Aa
1.9+0.3Aa
1.9+0.1Ba
ns
1.38
52
KoronadalH
igh
15
125
91.1+5.1Aa
56.1+5.4Ab
94.5+6.7Aa
0.001
3.95
61
1.7+0.2ABb
1.4+0.2ABb
3.1+0.2Aa
50.0001
16.91
40
Guimba
L
ow
13
81
38.8+3.7Bb
14.0+3.7Bc
76.0+13.4Ba
50.0001
6.78
54
1.4+0.2Bab
1.0+0.2Bb
1.7+0.2Ba
0.003
7.17
38
Calauan
L
ow
16
89
52.5+4.6Bb
22.2+4.9Bc
98.4+2.9Aa
50.0001
5.01
58
1.4+0.3Bb
1.0+0.1Bb
2.3+0.2Aa
50.0001
12.32
42
avg
68.0
35.8
90.4
1.6
1.3
2.3
P
50.0001
0.0001
0.006
0.006
0.008
50.0001
F
5.78
4.74
3.96
4.55
3.28
14.41
df
66
54
48
66
54
48
1Applicationsnotcountedintheseedbed.Inacolumn,means+SEMfollowedbyacommonupperca
seletterarenotsignificantlydifferent(P40
.05)byLSDtest.Inarow,means+SEMfo
llowedbyacommon
lowercaseletterarenotsignificantlydifferent(P40.05)byLSDtest.
2Fertiliserassubstituteforinsecticidewasincluded.
3Averageofinsecticide
usersonly.Doublespraysinresponsetoathresholdwascounted
astwoapplications.
52 J. A. Litsinger et al.
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
9/17
crotophos (33%), chlorpyrifos + fenobucarb (15%),
endosulfan (9%), methyl-parathion (8%), and cyper-
methrin (7%). All are broad spectrum materials and,
with the exception of methyl-parathion, did not differ
from those used in the threshold treatments. The
mean dosage per application of farmers ranged from0.20 to 0.23 kg a.i./ha for organophosphate, organo-
chlorine, and carbamate insecticides (Table IV). This
is half the recommended dosage which was set at the
minimum effective dosage level. Insecticide timing,
ranged among farmers from 20 to 26, 36 to 42, and
43 to 51 d.a.t. for the first through third applications
per crop. The late timing of the first application
showed the farmers practice likely would have
minimal effect against whorl maggot.
The calculation of mean dosage load is based on
multiplying the mean frequency of treated fields,
mean number of applications per treated field, and
mean dosage per application. Highest to lowestloads per crop over all sites were ranked from the
low value threshold 0.46 kg a.i./ha, farmers practice
0.45 kg a.i./ha, with the high value threshold at half
the load (0.21 kg a.i./ha) of the other two treatments
(Table IV). These data can be compared to the
prophylactic which was standardised at 1.3 kg a.i./
ha. Highest loads (40.6 kg a.i./ha) occurred in the
low threshold treatments of the high pest density
sites and in the farmers practice in Koronadal. In
order for the dosage loads of the threshold treat-
ments to be equal to those of farmers, Koronadal
farmers had to have sprayed twice as frequently percrop. This did not happen in Zaragoza where spray
frequency was similar, thus the low threshold load
was twice the farmers practice (0.35 kg a.i./ha).
Among the threshold treatments, significantly high-
er loads occurred in the high pest density than low
density sites reflecting the higher pest pressure. This
trend did not occur for the farmers practice. In
Guimba, because of the low frequency of fields
treated by farmers, there was no difference between
the farmers practice and the low value threshold. In
all sites, there were higher loads in the low threshold
level than high level (0.46 vs. 0.21 kg a.i.). Only in
Calauan was the farmers practice significantly
higher than the low threshold.
The pest control strategies were compared for
overall efficacy in controlling each pest group withthe full protection serving as a reference (Table V).
Prophylactic scored the highest overall control (41%)
among the practices which was significantly lower
than the full protection (60%) for all pests except
defoliators. Low and high thresholds were statisti-
cally similar (31 34%) to each other and the
farmers practice (24%) with only slight numerical
superiority in regard to whorl maggot and stem-
borers. Low threshold was similar to prophylactic for
all pests but leaffolders, while high threshold was
similar but for leaffolders and whorl maggot. Farm-
ers practice was only similar to prophylactic
regarding defoliator control.In terms of yield gain from control of each pest
group (comparing only those fields where thresholds
were reached), all practices were statistically lower
(206 342 kg/ha) than the full protection (729 kg/
ha) and, with the exception of stemborers, were
statistically undifferentiated. Yield gain from stem-
borer control by prophylactic was superior to that of
the farmers practice.
Further comparisons were made regarding total
yield over all fields. Comparing all sites and seasons,
the prophylactic was the only treatment equal to the
full protection (4.99 vs. 4.78 t/ha) (Table VI). Totalyield loss was 0.62 t/ha as the difference between the
full protection and the untreated. The prophylactic
treatment closed 66% of that yield gap. There was no
significant difference in yield between the prophy-
lactic, both threshold treatments, and farmers
practice (4.49 4.78 t/ha), but all were higher than
the untreated (4.37 t/ha).
The results from Calauan and Guimba mirrored
those of the overall analysis, but more treatment
separation occurred in the high pest density sites.
Table IV. Comparison of mean insecticide load per field between thresholds and farmers practice in four test sites, Philippines.
Dosage load per field (kg a.i./ha)1 Farmers insecticide
spray dosage2
Pest Low High Farmers (kg a.i./ha)
Site density Threshold threshold practice P F df (A)
Zaragoza High 0.72+0.25 A a 0.39+0.13 A b 0.35+0.17 B b 0.005 3.11 52 0.20
Koronadal High 0.62+0.21 A a 0.31+0.15 A b 0.67+0.27 A a 0.001 4.69 40 0.23
Guimba Low 0.22+0.08 B a 0.06+0.07 B b 0.28+0.09 B a 50.0001 5.09 38 0.22
Calauan Low 0.29+0.04 B b 0.09+0.06 B c 0.50+0.20 AB a 50.0001 7.28 42 0.23
avg 0.46 0.21 0.45 0.22
P 50.0001 50.0001 0.006
F 6.13 5.23 4.96
df 66 54 48
1Data averaged over all fields, derived from Table III by multiplying percentage fields treated by insecticide application frequency by the
dosage (0.4 kg a.i./ha used in thresholds and data in column (A) used for farmers. In a column, means+SEM followed by a common upper
case letter are not significantly different (P5 0.05) by LSD test. In a row, means+SEM followed by a common lower case letter are not
significantly different (P40.05) by LSD test. 2Dosage per application, synthetic pyrethroid insecticides not included.
Action thresholds for chronic rice insect pests 53
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
10/17
Averaging Zaragoza wet season yield data, all
treatments were significantly lower than the full
protection. Yields in both the threshold treatments
were equivalent to the prophylactic (all closing the
yield gap 47 50%), but the farmers practice did not
raise yield above the untreated. Dry season averageshad higher yield potential and in this case both the
prophylactic and low threshold were equal to the full
protection, closing the yield gap 90 and 69%,
respectively. The high threshold and farmers prac-
tice closed the yield gap 43 and 57%, respectively,
and were significantly higher yielding than the
untreated. With the first crop in Koronadal, only
the prophylactic equalled the full protection closing
the yield gap 81%. All the other treatments closed the
yield gap 20 29% but were statistically similar to the
untreated. The lower yielding second crop registered
a higher yield gap, and all treatments were superior to
the untreated but indistinguishable among them-selves closing the yield gap 23 49%.
Among sites Calauan recorded the most positive
marginal returns across treatments followed by
Zaragoza (Table VII). There was no trend based on
pest density. In Calauan the $48.10 marginal return
in the high threshold treatment was worth 376 kg of
unhusked rice. The two other sites had mostly
negative results. Among treatments the results were
highly varied by site with no clear best treatment
between the threshold treatments and farmers
practice (Table VII). The farmers practice had the
greatest mean benefit ($5.80) but only slightly higherthan the two threshold treatments. Low threshold
had the highest returns in Zaragoza but the results
were not significantly different than the high thresh-
old or farmers practice. Benefit cost ratios should be
2.0 or above to be attractive to farmers (Smith et al.
1989) which was only achieved in Calauan for both
threshold treatments. In no other treatment-site
combination did the benefit cost ratio exceed 1.5.
4. Discussion
4.1. Insect pests and densities
In this 13-year study in four prominent rice
growing areas in the Philippines, pest census and
yield loss measurement determined that whorl
maggot, defoliators, leaffolders, and stemborers were
the key chronic pests. Thresholds were surpassed for
at least one of the four pests in over three-quarters
(79%) of fields for each crop on average. These
figures include all the threshold characters tested but
would have been less if only the best characters had
been employed, lowering the should not have
treated error rate. With the exception of defoliators
in Calauan, all pests surpassed thresholds in eachstudy site at least once. Other rice pests rice bug,
whitebacked planthopper, and green leafhopper
occurred minimally; each surpassed thresholds in
only one site. The brown planthopper never reached
TableV.Comparisonof
correctiveactionstakenbetweenthreshold
sandotherpracticesdegreeofinsectcontrolandresultingyieldgain1.
Control(%)14WAT2
Damagedleaves
Stemborers
Yieldgain(kg/ha)3
Treatment
Whorlmaggot
Defoliators
Leaffolders
(deadhearts)
A
verage
Whorlmaggot
Defoliators
Leaffolders
Stemborers
Average
Fullprotection
51.7+4.0a
57.7+5.1a
80.1+4.4a
50.2+5.2a
59.9
821+68a
747+50a
742+55a
606+65a
729
Prophylactic
34.5+4.6b
48.1+5.1ab
57.5+4.4b
23.3+5.4b
40.9
397+78b
246+62b
489+57b
237+65b
342
Lowthresholds
26.4+3.9bc
43.7+5.0b
44.9+4.3c
18.9+5.3bc
33.5
215+66b
227+51b
327+55b
133+61bc
226
Highthresholds
22.1+2.8c
37.7+4.3b
43.9+3.5c
20.9+6.6bc
31.2
232+74b
221+35b
318+42b
115+51bc
222
Farmerspractice
16.0+4.1c
40.9+5.1b
31.2+5.0c
6.8+5.6c
23.7
254+78b
235+63b
322+66b
12+68c
206
P
50.0001
0.05
50.0001
50.0001
50.0001
50.0001
50.0001
50.0001
F
14.36
2.68
22.39
11.80
15.73
24.10
12.44
16.15
df
285
562
410
400
261
472
443
439
1Inacolumn,m
eans+SEM
followedbyacommonletter
arenotsignificantlydifferent(P40.05)byLSDtest.WAT=weeksaftertreatment.
2B
asedondamage.
3Dataonlyincludefields
wherethresholdsfor
eachrespective
pestwasexceeded.
54 J. A. Litsinger et al.
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
11/17
threshold levels in a single study field, mainly due to
resistant varieties adopted by most farmers.
The most ubiquitous pests were stemborers, whose
densities (2 3% damaged tillers over the entire
crop) and mode of damage, contributed more
significantly to yield loss than any of the other pest
groups. The same conclusion was reached by Savary
et al. (1994) but particularly in association with
weeds. Stemborer damage was measured in units of
the more important tillers rather than leaves. Eachmature tiller bears 10 18 leaves (Yoshida 1981),
thus for the same percentage, its damage is more
profound. Also a deadheart or whitehead is a severed
non-bearing tiller, while damaged leaves were rarely
entirely defoliated. In addition stemborer larvae
infest more tillers than are manifested as withered
tillers (unpublished data) further raising the percen-
tage of injury.
All four pest groups have been favoured by
inorganic fertilizer and irrigation, cultural practices
associated with modern varieties (Litsinger 1989).
Stemborers reputably have been the most important
insect pest of long-maturing, low-tillering, traditional
rices (Cendana and Calora 1967) but their impact
has been lessened by the earlier maturing, high-
tillering and narrow-stemmed modern rices. It was
surprising, therefore, that in Calauan, where many
farmers grow longer maturing varieties, that higher
stemborer damage did not occur. In Calauan the
120-day varieties would allow an extra (third)
stemborer generation to occur than is the case for
most modern rices. Natural enemies may have been
able to cope with this increase, whereas van der Goot
(1925) reported that with 6 9-month traditional
rices they usually could not.Whorl maggot and defoliators, not mentioned in
the pre-Green Revolution literature, emerged as new
pests. Whorl maggot, along with stemborers, were
the only pests that averaged damage levels above the
benchmark standards on a per-crop basis. Defoliators
and leaffolders are probably not normally a yield
threat in their own right due to the nature of the
damage (Heong 1990), but there is evidence that
defoliator injury can be significant when associated
with whorl maggot (Litsinger 1993). Both Zaragoza
and Koronadal had the highest combination of whorl
maggot and defoliators, and highest pest abundance
in general. Such densities may be related to extensive
irrigated rice areas (Loevinsohn et al. 1988), as bothsites lie in large irrigation systems.
4.2. Insecticide check method
Accurate yield loss estimates formed the basis of
threshold character evaluation. Unfortunately the
full protection treatment, which is the core of the
insecticide check method, only achieved the desired
480% control with leaffolders. Control was espe-
cially low in wet seasons due to monsoonal rainfall
which reduced insecticide residual activity. The
difference in control by season was most noticeable
in stemborers (least control in the wet season).
Typhoons were common in Luzon and when they
struck near harvest, the most vigorous growing
treatments (usually the full protection) were most
prone to lodge, biasing yield loss estimates down-
wards. Significant yield gain from thresholds, despite
high frequencies of correct decisions not to treat
scores, indicated that the insecticide check method is
not a good tool to match yield loss with a particular
pest in multipest crops such as rice.
A further confounding effect in trying to associate
losses with a single pest comes from synergistic losses
from multiple pests, each at subeconomic densities,that attack jointly causing significantly higher losses
than those caused by each acting singly (IRRI 1983,
1984; Wu et al. 1995). Furthermore the yield loss
contribution of each pest is significantly influenced
Table VI. Comparison of threshold treatments to other practices in terms of yield.
Yield (t/ha)1
Over all sites andZaragoza3 Koronadal3
Treatment Seasons2 Wet season Dry season First crop Second crop
Full protection 4.99+0.13 a 5.03+0.11 a 6.17+0.13 a 5.09+0.16 a 4.83+0.11 a
Prophylactic 4.78+0.13 ab 4.68+0.12 b 6.07+0.14 ab 4.98+0.16 ab 4.43+0.11 b
Low thresholds4 4.61+0.14 b 4.69+0.11 b 5.87+0.16 abc 4.67+0.16 bc 4.34+0.12 b
High thresholds5 4.49+0.14 b 4.70+0.11 b 5.62+0.13 c 4.62+0.16 bc 4.23+0.11 b
Farmers practice 4.53+0.15 b 4.28+0.13 c 5.76+0.14 bc 4.66+0.19 bc 4.46+0.11 b
Untreated 4.37+0.13 c 4.33+0.11 c 5.21+0.13 d 4.50+0.15 c 4.05+0.11 c
P 0.008 0.0002 0.003 0.007 50.0001
F 3.20 5.07 3.67 2.97 5.66
df 359 354 371 288 380
1In a column, means+SEM are not significantly different (P40.05) by LSD test. 2Data from 68 crops in Zaragoza, Guimba, Koronadal, and
Calauan sites. 3Data by fields, not crop averages. 4Low level characters, thus lower threshold values for a given character. 5High level
characters, thus higher threshold values for a given character.
Action thresholds for chronic rice insect pests 55
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
12/17
by associations of non-insect pest crop stresses
(Savary et al. 1994, 2000; Willocquet et al. 2000).
Correlations of insect pest densities to yield (damage
functions) therefore have been difficult to achieve in
rice (Litsinger et al. 1987; Litsinger 1991).
Large-scale field trials with 100-m
2
plot sizesemployed in the current study were dependent on
natural infestations and did not result in sufficiently
wide ranges of pest densities to derive damage
functions. Although total yield loss was high, it was
evenly distributed among growth stages, making
statistical distinctions between treatments difficult.
Therefore in future work, other yield loss assessment
methods should be considered. Artificial infestation
in units of individual hills can be readily achieved but
has had little success due to extreme yield variation
(Rubia et al. 1996). As soil fertility is highly variable
hill to hill in transplanted rice (Dobermann et al.
1995), larger 1 5-m2 units should tried (Litsinger1991).
4.3. Yield loss
Typhoon damage was prominent in Zaragoza and
the 1978 wet season (WS) crop was totally lost and
the data were not included as there was no way to
measure losses from insect pests. Included, however,
were years when significant typhoon damage oc-
curred (1980, 1985, 1988, 1989, 1990). Some of the
highest crop yields occurred, however, in dry seasons
following such losses (7.18 t/ha after the 1978 WSand 7.47 t/ha after the 1988 WS), the latter year
registering the highest yield per field of 9.47 t/ha.
Yield was enhanced as the naturally occurring
fertility that accumulates over time in flooded rice
soils (Cassman et al. 1996) was not removed in the
harvested wet season crop.
The most severe losses occurred when insect pest
damage was coupled with drought. Droughts
occurred in Guimba from the frequent shut down
of the deep well pumps as farmers could not pay
their electricity bills. Loss was particularly exacer-
bated in the very early maturing variety IR58
(maturity 585 d.a.t.) which severely constrained
crop compensation. Low losses in Calauan were
probably due to longer maturing varieties which
allowed greater compensation (Litsinger et al. 1987;
Litsinger 1993). Average field maturity for Calauan
cultivars was 119 days compared to 86 91 days at
other sites.
An important indirect outcome of this study was
the contribution of on-farm yield data for the
Philippines. As reported by Pingali et al. (1990),
yields have continued to increase relative to those on
research stations. Farmers yields, as determined
from crop cuts in socioeconomic surveys in CentralLuzon and Laguna, representing three of the study
areas (except Koronadal), increased from a mean of
2.19 t/ha in 1966 to 3.37 t/ha in 1979, a 53% increase
(Cordova et al. 1981). Farmers averaged 4.53 t/ha in
TableVII.Economicanalysisofthresholdtreatmentscomparedtootherpractices
1.
Marginalreturnfrominsecticide($
/ha)
2
Benefit:costratio
P
est
Low
High
Farmers
Site
density
Prophylactic
Lowthre
shold
Highthreshold
Farm
erspractice
P
F
df
Prophylactic
threshold
thres
hold
practice
Zaragoza
H
igh
711.30+9.80Bb
11.40+10
.20Ba
71.70+3.20Bab
3.50
+5.70Bab
0.007
4.23
52
0.9
1.2
1.0
1.1
KoronadalH
igh
733.70+29.60Cb
717.60+13
.50Cab
720.90+24.90Cab
79.10
+5.50Ba
0.005
3.29
40
0.6
0.6
0.5
0.8
Guimba
Low
734.40+41.40Cb
722.90+12
.00Cb
77.80+9.60Ba
0.50
+0.40Ba
50.0001
2.99
38
0.6
0.4
0.7
1.0
Calauan
Low
24.10+20.50Ab
43.30+42
.60Aa
48.10+27.10Aa
28.10
+23.40Ab
50.0001
5.12
42
1.3
2.1
2.6
1.5
avg
713.80
3.58
4.41
5.76
0.8
1.1
1.2
1.1
P
50.0001
0.004
50.0001
0.003
F
4.36
6.34
5.99
3.58
Df
66
66
54
48
1Inacolumn,m
eans+SEM
followedbyacommonuppercaseletterarenotsignificantlydifferent(P40.05)byLSDtest.Inarow,means+SE
M
followedbyacommonlowercaseletter
arenotsignificantly
different(P40.05)byLSDtest.
2Costbasisbasedonpricesin1986forinsecticide,unmilledrice$0.128/kgfarmgate,interestonmaterials60%perseason,labour8htospray1ha,labour
$0.10/h,interestfor
labour33%perseason,pestmonitoring60hperseasonforthresholdsand4hperseasonforfarme
rsmonitoring.
56 J. A. Litsinger et al.
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
13/17
the current study, a further 34% increase and double
the 1966 pre-Green Revolution average per crop.
Dry season yields were 12% higher than the wet
season (excluding Koronadal which lies south of
monsoonal influence) based on increased solar
radiation (Yoshida 1981). Monsoon clouds particu-larly at grain filling stages can severely limit yield
potential. Farmers responded to the higher potential
in the dry season by increasing N rates, particularly in
Luzon, where farmers N averaged 68 kg/ha in the
wet season and 101 kg/ha in the dry season, a 49%
increase (average from Guimba, Zaragoza, and
Calauan sites). Added N has less effect in Mindanao
where soil fertility is naturally high (farmers averaged
35 kg N/ha with no difference between seasons).
Differences in crop management, yield potential,
and insect pest incidence can explain the differences
between sites in the crops ability to compensate from
pest injury. High compensation was observed in bothGuimba and Calauan where pest incidence was
generally low and N inputs high. In Zaragoza under
high pest pressure, high compensation occurred
during the dry season, whereas in the wet season,
the crop could not outgrow damage. In Koronadal
pest incidence was high and compensation was not
recorded in any crop probably as added N levels were
too low.
4.4. Nitrogen substitution
The best threshold characters predicted both yieldloss and damage benchmarks 490% of occasions
(Litsinger et al. 2005). The constraint with AT
performance was not in choice of characters but was
due to the poor result of the insecticide response. N
substitution attempted a new strategy to tap the
crops resilience to pest damage. The property of
modern rices to compensate for high levels of pest
damage has been overshadowed by the accomplish-
ments in genetic resistance (Khush 1989), but both
act in a complementary fashion. The key to resilience
in modern rices over that of traditional rices is their
high tillering ability. Thus if a tiller is severed by
stemborers, neighbouring ones can compensate
(Rubia et al. 1996). Modern rices also produce
many more spikelets per area than traditional types
and thus have a larger physiological sink. When a
whitehead occurs the photosynthate can transfer to
unfilled spikelets of an adjacent tiller. Trials showed
compensation is enhanced by optimal crop manage-
ment with N application at the top of the list
(Litsinger 1993). Modern rices also do not lodge as
easily as the tall traditional types in response to
higher N rates.
N substitution supplements a technology farmers
already use (but not necessarily optimally) and doesnot have the negative effects of insecticides in terms
of safety, environmental hazards, and impacting
beneficial arthropods (Pingali and Roger 1995).
The N response, however, produced mixed results.
The applied N rate of 25 kg/ha has the potential to
increase yield 500 kg/ha (Yoshida 1981), but sig-
nificant yield gains only ranged from 220 to 389 kg/
ha, less than the potential. The additional N was
applied when a pest threshold was reached which
may not always have been the most physiologicallyappropriate time.
N use is a double-edged sword as overuse
promotes pest abundance, particularly diseases (Sa-
vary et al. 1995). Most farmers apply 70% of total N
11 21 d.a.t., and thus overstimulate plant to plant
competition and disease susceptibility. N broadcast
into paddy water is lost within several weeks of
application, and as a result only about 40% enters the
crop (Cassman et al. 1996). Thus more frequent
applications are generally required. Applications in
later stages favour compensation from leaffolder and
stemborer damage by delaying leaf senescence (Peng
et al. 1996) but also prolongs pest attack.
4.5. Threshold treatments compared to other practices
The prophylactic treatment of three applications
was designed to prevent damage from early season
whorl maggot and defoliators as well as stemborer
whiteheads. The total insecticide dosage of 1.3 kg
a.i./ha was three times that of the low threshold and
farmers practice. Due to its higher cost and only a
slight yield benefit compared to other treatments, it
was the lowest performing treatment economically.
Low threshold resulted in about two-thirds of thefields being treated with an average of 1.6 applica-
tions per crop (half that of the prophylactic). Its
efficacy levels were statistically similar to the prophy-
lactic against all pests but leaffolders. Yield gain was
similar to prophylactic but equal to the other
practices as well. Total yield and efficiency in
insecticide usage were equal in all crops to the
prophylactic giving better economic returns.
The high threshold treatment resulted in only one-
third of all fields receiving insecticide with an average
of 1.3 applications each for a low dosage load of
0.21 kg a.i./ha, half that of the low threshold
treatment and one-sixth that of the prophylactic.
Efficacy against pests suffered as a result, as its
percentage control was equal to the prophylactic in
only two pest groups (defoliators and stemborers).
But yield gain was similar to the prophylactic. Total
yield was equal to the prophylactic in most crops. Its
economic returns were 22% better than the low
threshold and four times better than the prophylactic.
Overall the economic returns were similar to the low
threshold in being acceptable in only one site.
The farmers practice applied insecticide to a
similar percentage of fields as the low threshold
treatment in the high pest sites (490%). The dosageload (0.45 kg a.i./ha) was equal to the low threshold
because of the high application frequency over all
sites (2.3 applications) despite the lower dosage per
application. Filipino farmers chronically underdose
Action thresholds for chronic rice insect pests 57
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
14/17
(Litsinger et al. 1980b, 1982; Marciano et al. 1981;
Pineda et al. 1984; Waibel 1986), and resist
extension efforts to increase as they experienced
dizziness and headaches upon doing so (Goodell et
al. 1982). They resist increasing spray volume as they
believe the crop is damaged from excessive walkingthrough the field, although research shows otherwise
(Arida and Shepard 1987). By increasing only the
concentration in the spray tank and spraying directly
in front while walking, they have increased exposure.
They believe insecticides act in the same way as
fertilizers in that a small amount will produce some
response. Thus they do not understand the nonlinear
dosage mortality relationship (Litsinger et al. 1980b).
As a result of low dosages by farmers, the quantity of
insecticide applied per crop was similar between the
threshold treatments and the farmers practice.
The threshold method of weekly crop monitoring
is designed to respond to variations in pest density,whereas farmers were often guided by a combination
of prophylactic insecticide usage (timed to fertilizer
application) and use of very low threshold levels such
as the mere presence of flying moths (Bandong et al.
2002). In Calauan, farmers applied at frequencies
more consistent with a high pest density site. That
site is urbanised and consequently visited more
frequently by chemical company representatives.
Compared to the high threshold treatment, farm-
ers applied twice the dosage load per field with
similar efficacy, yield gain, and total yield. The
reason farmers insecticide practices of frequent,albeit low dosage applications, gave relative good
yield responses is unknown. The average 0.22 kg a.i./
ha dosage is just above threshold of the dosage
mortality curves of most insecticides, generally
causing 20 30% mortality (unpublished dosage trial
data). This meant that about half of their applications
were sublethal, but testing these dosages in the
laboratory, Tevapunchom and Heong (1991) noted
subtle negative effects on leaffolder development.
Sublethal doses may have affected the adult stages of
pests, which were not examined.
Farmers practice outgained the high threshold
treatment by 24% in economic returns on average
with the exception of Calauan. The benefit cost ratio
was similar to both threshold treatments but did not
meet the economic standards in any one site. The
fact that the farmers practice was uneconomical
supports the evidence that Filipino farmers use
insecticide to reduce risk of crop failure rather than
optimising profit (Waibel 1986).
The largest differences between threshold treat-
ments and the farmers practice, were timing of
application, dosage per application, and in the low
pest density sites, percentage fields treated within a
crop. On the other hand, similarities existed betweenthresholds and farmers practice in choice of in-
secticides, and in the high pest density sites with low
level thresholds, percentage of fields treated and
number of applications per field.
4.6. The future of action thresholds
It was hypothesised that threshold characters based
on life stages (eggs, larvae, adult moths) rather than
damage would improve insecticide timing from
earlier warning (Litsinger et al. 2005). This turnedout not to be true as the most effective characters,
with few exceptions, were based on insect damage.
The most likely reason for the poorer showing of
insect stages was the profound effect of natural
enemies (Ooi and Shepard 1991). Those characters
further removed in development time, such as moths,
from damaging larval stages proved least reliable
probably because of the greater time for natural
enemy activity. Stemborer egg mass characters had
incorporated the contribution of egg parasitoids into
the threshold but did not account for egg predators
which can be equally effective. Whorl maggot eggs
were the exception, producing good results, as itscolonisation generally precedes that of natural
enemies (van den Berg et al. 1988).
Both farmer-inspired threshold characters: (1)
flushed moths and (2) earlier-planted fields produced
mixed results. Moths turned out to be poor
monitoring tools for both leaffolders and stemborers,
for the reason just mentioned, resulting in low
frequencies of correct decisions and high error rates.
Characters employing monitoring of earlier-planted
fields showed promise with whorl maggot and
defoliators but not with the other two chronic pests.
In the case of whorl maggot better control was relatedto application early in the crop.
Because of the high degree of crop compensation
inherent in high tillering and longer-maturing vari-
eties, crop management should play a central role in
IPM along with conservation of natural enemies.
Despite great adoption of modern varieties, there lies
great challenges ahead to improve crop management
as illustrated by normally wide yield ranges in a given
community, from 51 to 4 9 t/ha; measured in this
as well as other studies (Pingali et al. 1990). In the
latter study, farmers were grouped by yield, with the
top one-third achieving yields on a par with research
stations. Thus the current yield gap and rationale for
extension efforts is not between farmers and
researchers, but between the top one-third and lower
two-thirds of farmers. The top one-third is achieving
maximum pest compensation benefits. But the
relationship of IPM vis-a-vis crop management
practices is complex due to two opposing forces:
(1) the great capacity of high tillering and longer-
maturing rices that bolster compensation from
damage counterbalanced by (2) the synergistic effect
of multiple stresses in reducing yield, with each pest
being just one stress (Litsinger 1991).
In the monsoon season in Zaragoza (Figure 1,Table VI), due to later planting compared to
Guimba, the crop is usually in the grain filling stage
during the main typhoon season. The effect is to
reduce the effectiveness of insecticide due to frequent
58 J. A. Litsinger et al.
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
15/17
rains but farmers select longer maturing varieties
which have longer periods of compensation. Yield is
limited due to monsoonal clouds. Thus it is doubtful
that improved crop management, other than select-
ing longer maturing varieties, will bring higher
returns. Risk of crop failure is too high for farmersto take advantage of applying higher N rates. Farm-
ers, therefore, should strive for earlier planting
(Savary et al. 1994). The lower yields in the wet
season can be made up in the dry season. This is
contrasted with the dry season with greater yield
potential where the yield gap was wider and
compensation was not lacking from inadequate N
usage but from the farmers selection on earlier
maturing varieties due to the limited water supply.
In Koronadal it may be worthwhile for farmers to
grow longer-maturing varieties, apply higher N rates,
or practice better weed management. In contrast
Calauan, a low pest density site, which has fewer cropstresses appeared to offer the best conditions for use
of thresholds. Farmers already use high levels of N
and good crop management in general. The agro-
nomic feature that distinguishes Calauan is that
farmers cultivate long maturing varieties and use of
the dapog seedbed method. N usage averaged 73 and
82 kg /ha in the wet and dry seasons. These compare
well to Guimba 77 and 112 days. The benefit in
compensation from longer-maturing varieties has
already been shown (Litsinger et al. 1987; Litsinger
1993). Dapog seedbed minimizes transplanting
shock and may have helped compensate for whorlmaggot damage.
Threshold levels will need to be locally fine tuned
based on experience and current risk assessment: the
better the management, the higher the thresholds.
Risk may change from season to season even for the
same farmer (Litsinger 1993). Leaf colour charts
could be used to determine a crops N demand
(Singh et al. 2002). If there are several stresses, say
insect pests, in a given growth stage, then only one of
them may need to be confronted, while leaving the
more difficult-to-control to crop compensation.
Acknowledgements
We are duly appreciative of the generous coopera-
tion provided by over 400 farmers in the study sites.
Their willingness to become experimenters with the
research teams and devote at times a tenth of their
ricelands to trials is a testament to their desire to seek
improvements in rice production technology. Many
locally hired project staff were responsible for
conducting the trials and their invaluable contribu-
tions are acknowledged. Those assisting in Zaragoza
were Catalino Andrion and Rodolfo Gabriel, in
Guimba George Romero, in Calauan MarianoLeron, Eduardo Micosa, and Carlos de Castro, and
in Koronadal Hector Corpuz, Joseph Siazon, Beatriz
Velasco, and Anita Labarinto. Cooperation of the
staff in the Central Luzon and Mindanao regions of
the Philippine Department of Agriculture is highly
appreciated.
References
Arida GS, Shepard BM. 1987. Walking the rice paddy for pest
sampling does not affect yield. International Rice Research
Newsletter 12(3):33.
Bandong JP, Canapi BL, dela Cruz CG, Litsinger JA. 2002.
Insecticide decision protocols: a case study of untrained Filipino
rice farmers. Crop Protection 21:80316.
Cassman KG, Gines GC, Dizon MA, Samson MI, Alcantara JM.
1996. Nitrogen-use efficiency in tropical lowland rice systems:
contributions from indigenous and applied nitrogen. Field
Crops Research 47:112.
Cendana SM, Calora FB. 1967. Insect pests of rice in the
Philippines. In: Major insect pests of the rice plant. Baltimore,
MD, USA: Johns Hopkins Press. pp 591616.
Cordova V, Papag A, Sardido S, Yambao LD. 1981. Changes in
practices of rice farmers in Central Luzon: 196679. In: 12th
Annual Scientific Meeting of the Crop Science Society of the
Philippines, Bacnotan, La Union 2224 April 1981. p 23.
Department of Entomology, International Rice Research Institute
1985. Insecticide Evaluation Report for 1984. Los Banos
Philippines: IRRI. p 232. (tables 136138, figures 1423).
Department of Entomology, International Rice Research Institute
1988. Insecticide Evaluation Report for 1986. Los Banos,
Philippines: IRRI. p 355. (tables 92103, figures 753).
Dobermann A, Pampolino MF, Neue HU. 1995. Spatial and
temporal variation of transplanted rice at the field scale.
Agronomy Journal 58:71220.
Dyck VA, Htun Than, Dulay AC, Salinas GD, Orlido GC. 1981.
Economic injury levels for rice insect pests. Agricultural
Research Journal of Kerala 19:7585.
Goodell G. 1984. Challenges to integrated pest managementresearch and extension in the Third World: Do we really want
IPM to work. Bulletin of the Entomological Society of America
30:1826.
Goodell GE, Kenmore PE, Litsinger JA, Bandong JP, dela Cruz
CG, Lumaban MD. 1982. Rice insect pest management
technology and its transfer to small-scale farmers in the
Philippines. In: Report of an exploratory workshop on the role
of anthropologists and other social scientists in interdisciplinary
teams developing improved food production technology. IRRI
and the Division for Global and Inter-regional Projects. Rome,
Italy: United Nations Development Programme. pp 2541.
Heinrichs EA. 1986. Perspectives and directions for the continued
development of insect-resistant rice varieties. Agriculture,
Ecosystems, and the Environment 18:936.
Heinrichs EA, Pathak PK, Dyck VA, Chelliah S, Saxena RC,Litsinger JA. 1978. Guide to the management of insect pests of
lowland rice in tropical Asia. In: FAO short course on integrated
pest control for irrigated rice in south and southeast Asia,
October 16November 14, 1978. Manila, Philippines: Bureau of
Plant Industry.
Heong KL. 1990. Feeding rates of the rice leaffolder, Cnaphalo-
crocis medinalis (Lepidoptera: Pyralidae), on different plant
stages. Journal of Agricultural Entomology 7:8190.
Heong KL, Escalada MM. 1997. Perception change in rice pest
management: A case study of farmers evaluation of conflict
information. Journal of Applied Communications 81:317.
International Rice Research Institute (IRRI) 1983. Insect combi-
nations. In: Annual Report for 1982. Los Banos, Philippines:
IRRI. pp 2023.
International Rice Research Institute (IRRI) 1984. Yield lossescaused by multiple species infestations. In: Annual Report for
1983. Los Banos, Philippines: IRRI. pp 1878.
International Rice Research Institute (IRRI) 1985. International
Rice Research: 25 Years of Partnership. Los Banos, Philippines:
IRRI. p 188.
Action thresholds for chronic rice insect pests 59
8/2/2019 Yield Loss & Action Thresholds of Chronic Insect Pests
16/17
Kenmore PE, Litsinger JA, Bandong JP, Santiago AC, Salac MM.
1987. Philippine rice farmers and insecticides in thirty years of
growing dependency and new options for change. In: Tait J,
Napompeth B, editors. Management of pests and pesticides:
farmers perceptions and practices. Westview Studies in Insect
Biology. London, UK: Westview Press. pp 98108.
Khush GS. 1989. Multiple disease and insect resistance for
increased yield stability in rice. In: Progress in irrigated rice
research. Los Banos, Philippines: IRRI. pp 7992.
Litsinger JA. 1989. Second generation insect pest problems on
high yielding rices. Tropical Pest Management 35(3):23542.
Litsinger JA. 1991. Crop loss assessment in rice. In: Heinrichs EA,
Miller TA, editors. Rice insects: management strategies. New
York: Springer-Verlag. pp 165.
Litsinger JA. 1993. A farming systems approach to insect pest
management for upland and lowland rice farmers in tropical
Asia. In: Altieri MA, editor. Crop protection strategies for
subsistence farmers Westview Studies in Insect Biology,
Boulder, CO, USA: Westview Press. pp 45101.
Litsinger JA. 1994. Cultural, mechanical, and physical control of
rice insects. In: Heinrichs EA, editor. Biology and management
of rice insects. New Delhi, India: Wiley Eastern Ltd. pp 54984.
Litsinger JA, Lumaban MD, Bandong JP, Pantua PC, Barrion AT,
Apostol RF, Ruhendi 1980a. A methodology for determining
insect control recommendations. IRRI Research Paper Series
No. 46:131.
Litsinger JA, Price EC, Herrera RT. 1980b. Small farmer pest
control practices for rainfed rice, corn, and grain legumes in
three Philippine provinces. Philippine Entomologist (1978)
4:6586.
Litsinger JA, Canapi B, Alviola A. 1982. Farmer perception and
control of rice pests in Solana, Cagayan Valley, a pre-green
revolution area of the Philippines. The Philippine Entomologist
5:37383.
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 lossfrom insect pests in wetland and dryland environments of Asia
with emphasis on the Philippines. Insect Science and its
Application 8:67792.
Litsinger JA, Bandong JP, Canapi BL, dela Cruz CG, Pantua PC,
Alviola AL, Batay-An III E. 2005. Evaluation of action thresh-
olds for chronic rice insect pests in the Philippines: II. Whorl
maggot and defoliators. International Journal of Pest Manage-
ment.
Loevinsohn ME, Litsinger JA, Heinrichs EA. 1988. Rice insect
pests and agricultural change. In: Harris MK, Rogers CE,
editors. The entomology of indigenous and naturalized systems
in agriculture. Boulder, CO, USA: Westview Press. pp 16182.
Marciano VP, Mandac AM, Flinn JC. 1981. Insect management
practices of rice farmers in Laguna. Philippine Journal of Crop
Science 6:1420.Matteson PC. 2000. Insect pest management in tropical Asian
irrigated rice. Annual Review of Entomology 45:54974.
Matteson PC, Gallagher KD, Kenmore PE. 1994. Extension of
integrated pest management for planthoppers in Asian irrigated
rice: empowering the user. In: Denno RF, Perfect TJ, editors.
Planthoppers: Their ecology and management. New York:
Chapman & Hall. pp 65685.
Miyashita T. 1985. Estimation on the economic injury level in the
rice leafroller Cnaphalocrocis medinalis Guenee (Lepidoptera
Pyralidae). I. Relation between yield loss and injury of rice
leaves at heading or in the grain filling perod. Japanese Journal of
Applied Entomology and Zoology 29:736.
Morse S, Buhler W. 1997. Integrated Pest Management Ideals
and Realities in Developing Countries. London: Lynne Rienner
Publ. p 166.Norton GA, Mumford JD. 1993. Decision tools for pest manage-
ment. London: CAB International.
Ooi PAC, Shepard BM. 1991. Predators and parasitoids of rice
insect pests. In: HeinrichsEA editor. Biology and management of
rice insects. New Delhi, India: Wiley Eastern Ltd. pp 5845612.
Peng S, Garcia FV, Laza RC, Sanico AL, Cassman KG. 1996.
Increased N-use efficiency using a chlorophyll meter on high-
yielding irrigated rice. Field Crops Research 47: 24352.
Pineda R, Duff B, Heinrichs EA, Carbonell P. 1984. Insect control
practices on irrigated and rainfed rice farms in Nueva Ecija,
Philippines. In: The consequences of small rice farm mechan-
ization project. Los Banos, Philippines: IRRI Working Paper
No. 102. pp 120.
Pingali PL, Roger PA. 1995. Impact of pesticides on farmer health
and the rice environment. Boston, USA: Kluwer Academic
Publishers. p 664.
Pingali PL, Moya PF, Velasco LE. 1990. The post-green
revolution blues in Asian rice production. IRRI Social Science
Division Papers, No. 90. pp 133.
Pingali P, Hossain M, Gerpacio RV. 1997. Asian rice bowls: The
returning crisis? Oxon, UK: CAB International. p 341.
Reissig WH, Heinrichs EA, Litsinger JA, Moody K, Fiedler L,
Mew TW, Barrion