and also two in Louisiana in 1988. Lower dosage
rates of methyl parathion were applied in the second
test site. Insecticides evaluated included Ambush
(permethrin), Asana (esfenvalerate), Baythroid (cy-
fluthrin), Cymbush (Cypermethrin), Karate (Iambda-
cyhalothrin), methyl parathion or
micro-encapsulated methyl parathion, Orthene
(acephate), Scout (tralomethrin), and an untreated
control.
All plots were sampled weekly using standard
15-inch diameter sweep nets (Kogan and Pitre), and
treatments were applied whenever stink bug popula-
tion densities reached the treatment threshold of six
per 25 sweeps during soybean growth stages R∣
(pods developing) through Rβ (full green bean devel-
oped in the pod) (Adams and McPherson). Approxi-
mately 30 days separate Ri from R«. AU insecticides
were applied on the same date to control a uniform
distribution of stink bug population densities that
exceeded the treatment threshold in all plots. These
single insecticide applications provided season-long
stink bug control in all plots each year, except for
Georgia in 1988. In the 1988 Georgia experiment, a
second insecticide application was necessary for all
plots two weeks after the first treatment to maintain
stink bugs below the threshold level. For all other
years and locations only one application was ap-
pUed.
Although it was the objective of this study to wait
for an economic threshold level, this never occurred
in Florida in 1989, so applications were made at
one-half the threshold. In practice, soybean produc-
ers often only partially adopt threshold recommen-
dations. They may apply insecticides at a
sub-economic threshold level, concerned that dam-
age wiU occur if they wait too long. In soybean
production, a discussion of the feasibility of partiaUy
adopting economic thresholds, under risk, is pre-
sented in Szmedra et al.
No distinction was made between the Bragg and
Braxton soybean varieties in this study, because ear-
lier reports documented no differences between cul-
tivars in the same maturity group (GUman et al.). All
plots were harvested with a small plot combine with
yield and seed quality evaluations conducted. Four
100-seed samples were randomly selected from each
treatment. Using criteria reported by Jenson and
Newsom, these seeds were manually categorized as
having either Ught, moderate, heavy, or no stink bug
damage according to their appearance. Light dam-
age indicates seeds with little damage, moderate
damage refers to shrivelled and discolored seeds, and
heavy damage indicates severely shrivelled and de-
formed seeds.
Seed QuaUty Adjustment
Based on an elevator’s usual practice of dockage,
seeds with no damage or light damage are catego-
rized as seeds without damage, and seeds with mod-
erate or heavy damage are categorized as damaged
seeds. No dockage is appUed to seeds without dam-
age. For damaged seeds, only one-fourth of actual
damage is counted for dockage, because damage
involving discoloration and wrinkled surface usuaUy
w∏l not hurt the oil and protein content of the seeds.
For damage below eight percent (equivalent to 32
percent of actual damage), each one percentage point
damage is docked two cents per bushel. For damage
beyond eight percent each additional 0.5 percentage
point damage is docked three cents per bushel. Soy-
bean price before dockage, $5.96 per bushel, is the
average October soybean price received by Georgia
producers from 1983 to 1989 (Georgia Crop Report-
ing Service).
CONCEPTUAL MODEL
The stochastic economic state variable is annual
per acre profit, π⅛ for field experiment k, across
stations and years, and insecticide j.
(1) πkj = YltjP(l - Dkj) - [A(r.j + v) + C](l + i)
-NC-L,
where Y⅛ denotes stochastic yield in bushels per
acre; P and Dkj are per-bushel soybean price and
stochastic price reduction, dockage, for soybean
damage, respectively. Total cost per acre is the sum
of cash costs and noncash costs. Cash costs can be
divided into cost of insecticides, A(r.j + v)(l + i), and
all other cash costs, C(l + i). Cost per acre of
insecticide, A(rj + v)(l + i), is determined by the
number of appUcations, A, times the sum of per acre
cost for insecticide j, rj, and per unit cost of applica-
tion, v, multiplied by (1 + i), where i is the biannual
interest rate. A six-month loan is assumed. In all
experiments A = 1, except for the Georgia experi-
ments in 1988 where A = 2. Other cash costs include
seed, fertilizer, herbicides, scouting, machinery, ma-
chinery taxes, land rent, and interest on operating
capital. Noncash costs, NC, include depreciation,
average investment, and housing of machinery, and
L denotes cost of unpaid family labor.
Costs
Insecticide costs, for the alternative chemicals,
were based on the unit prices of active ingredients
from a representative southeastern agricultural
chemical supply company. The costs of insecticides
85