See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233498446

Evaluating Single Treatment Data Using Abbott's Formula With Reference to

Insecticides

Article  in  Journal of Economic Entomology · December 1985

DOI: 10.1093/jee/78.6.1179

CITATIONS READS

160 14,257

2 authors, including:

Arthur Retnakaran

130 PUBLICATIONS   3,540 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Insect Molecular Biology View project

All content following this page was uploaded by Arthur Retnakaran on 16 May 2014.

The user has requested enhancement of the downloaded file.

21706 Fleming, R. A. 1985
Evaluating Single Treatment Data ^Jtng ^fcbotfi Formula
With Reference to Insecticides

RICHARD FLEMING AND ARTHUR RETNAKARAN

Forest Pest Management Institute, Canadian Forestry Service,

P.O. Box 490, Sault Ste. Marie, Ontario P6A 5M7, Canada

FORUM: J. Econ. Entomol. 78: 1179-1181 (1985)

ABSTRACT Abbotts formula is often applied to entomological field data to distinguish
the effects of pesticide treatment from those caused by natural factors. Although one would
normally hope to use inferential statistics in applying Abbott's formula, logistical and eco
nomic constraints sometimes preclude the necessary experimental replication. We show that
the use of Abbott's formula in experiments involving a single treatment and a single check
plot usually produces a biased estimate of the population reduction due to treatment. The
reasons for this bias are presented as guidance in the design of pesticide trials and in the
interpretation of their results.

To adhere rigidly to the strict requirements of Rearranging equation 1 produces Abbott's (1925)
inferential statistics, one must obtain data from formula for the population reduction due to treat
experiments that are adequately replicated. In ment:
many types of experiments, especially those in
volving field trials, such replication is not always R = 1 - ajac where 0 < act 0 < ar (2)
possible. For example, Moffitt and Westigard (1984) (Here R < 0 implies a population increase, or neg
noted in their report on the suppression of the ative decrease, due to treatment.) This equation
codling moth population in pear orchards by mat describes a curve by which, for a fixed value of
ing disruption with sex pheromone that "No with- the total per capita change, a„ R increases mono-
in-orchard treatment replication was economically tonically (from R « —oo when ac « 0) at a decel
or, in most cases, physically possible/' Hurlbert erating rate as natural per capita change, ac, in
(1984) stressed this point in his monograph on the creases. For ac ^> an the population reduction due
design of ecological field experiments: "Replica to treatment, R, approaches 1 asymptotically from
tion is often impossible or undesirable when large- below and is little affected by ac.
scale systems (whole lakes, watersheds, rivers, etc.) The population reduction, R, is widely used as
are studied. When gross effects of a treatment are a measure of the impact of pesticide application
anticipated, or when only a rough estimate of ef above and beyond the impact of naturally occur
fect is required, or when the cost of replication is ring factors (which also influence the dynamics of
very great, experiments involving unreplicated treated populations and presumably would have
treatments may also be the only option." Examples occurred anyway). Thus, Abbott's formula is used
of such trials can be readily seen in the literature to distinguish treatment effects from the effects of
(Overhulser et al. 1980, Morris 1984, Bostanian et naturally occurring factors (Sower et al. 1982).
al. 1985, Nord et al. 1985, Philogene et al. 1985, The per capita rates of change are generally
Retnakaran and Grant 1985, Solomon 1985, Yous- estimated as:
sef and McManus 1985).
In each of these examples, the impact of a par at = T2/T1,
ticular treatment (e.g., chemical) on a target (e.g.,
pest) population is measured quantitatively in terms and (3)
of the change in population density beyond that ac = C2/C1,
which would have occurred in the absence of the
treatment. For instance, if ac is the naturally oc where Tl and T2 are the pre- and posttreatment
curring per capita change in population density as population densities in the treatment plot, and CI
measured in the check plot (and assumed to be the and C2 are the corresponding check plot popula
same as in the treatment plot if treatment did not tion densities. Using equation 3 in equation 2, Ret
occur), then the total (natural plus treatment in nakaran (1980) arrived at a computationally con
duced) per capita change measured in the treated venient form of Abbott's formula:
plot is
R = 1 - (T2 x C1)/(T1 x C2). (4)
at = ac - acR = ac{\ - R), (1)
where —aeR represents the treatment effects on The same formula can be descriptively expressed
per capita change. in terms of percentage as:
 /posttreatment pretreatment •
population population
in treatment in check
population = 1 - x 100 (5;
pretreatment posttreatment
reduction
population population
in treatment in check

Implicit in the use of Abbott's formula and in Furthermore, we might assume that the relative
the selection of the plots is the assumption that the error or bias in estimating posttreatment popula
ecological influences on the check and treatment tion densities is the same on the control plot and
populations are identical at the time of treatment. the treatment plot (so T2*/T2 = C2*/C2 where
Hence, since population size generally affects both T2* and C2* are the treatment and check post-
natural rates of change (e.g., via competition, pre- treatment density estimates, respectively). Using
dation, parasitism, disease) and dispersal (cf. Krebs equation 4 to write the difference between the true
1972), ideally the treatment and check plots should or actual population reduction, R, and its estimate,
have equal pretreatment densities. In practice, R*, results in the equation
however, it may not always be possible to select
R-R* = [(C1*/T1*) - (Cl/Tl)] (T2/C2).
plots with identical population densities. When
several treatments are considered, it is customary Since CI = Tl, substituting for Z with equation 6
to select many plots in the hope that they can be results in the error or bias in estimating population
matched with treatment plots of similar popula reduction of
tion density (e.g., Retnakaran 1981). When several
R -R* = (Z -1)(T2/C2).
check plots of varying population densities are
available, a curve of the natural per capita rate of This is the equation of a line of slope (T2/C2) and
change for the population densities of concern can intercept zero in the (R - fl*) x (Z — 1) plane.
be constructed and the relationship obtained can This equation shows that the magnitude of the
be used to isolate treatment effects from ecological estimation error (or bias), R - R*, for population
influences (G. Howse, personal communication). reduction depends on the ratio of posttreatment
However, economic or logistical constraints or both densities, (T2/C2), and on the magnitude of the
often limit the trial to one treatment plot and one deviation of the ratio of pretreatment estimates
check plot (e.g., Retnakaran and Grant 1985). The from one, Z — 1. More important, use of the sam
question then arises: should the plot with the great ple plot with the larger pretreatment density es
er pretreatment estimate be selected for the check timate for the check (so Z > 1) underestimates
or the treatment? population reduction (i.e., R* < R); using the oth
When the pretreatment estimates are not equal, er plot for the check (so Z < 1) produces an over
two alternative situations are possible: the actual estimate.
pretreatment population densities may also be dif This problem is illustrated by the data shown in
ferent, or the actual pretreatment densities may Table 1. In trial 1, the prespray density in the
be equal and only their estimates unequal. For the check plot is identical to that in the treatment plot,
first situation, in the absence of evidence to the which gives an unbiased estimate of population
contrary, we can assume the ecological generality reduction of 80%. In trials 2 and 3, the change in
that net per capita change through the combined the pre- and postspray levels in the check are pro
effects of natural mortality, reproduction, and dis portional and, therefore, do not alter the result.
persal usually decreases as population density in Such a situation, however, cannot be predicted at
creases (e.g., Pielou 1969). Then, calculating the the time of the spray operation. Trials 4 and 5
natural per capita change from the plot with the represent the more probable situation where the
larger pretreatment population will usually un postspray check densities show a greater decrease
derestimate ac in the other (treatment) plot, and
hence underestimate R in equation 2. Doing the
reverse leads to an overestimate of the population Table 1. Example of data for a field experiment
reduction due to treatment.
The second situation, where the pretreatment Population in Population in % popu
densities are actually equal (i.e., CI = Tl) al Spray treatment check lation
though their estimates are not, leads to the same trial Pre Post Pre Post reduc
conclusion. For instance, let the ratio of the esti spray spray spray spray tion

mates of the pretreatment densities be 1 100 10 100 50 80

2 100 10 50 25 80
Z = C1*/T1* (6)
3 100 10 150 75 80
4 100 10 50 50 90
where CI* and Tl* denote the check and treat
5 100 10 150 50 70
ment pretreatment density estimates, respectively.
 Flkminc; and Rktnakara.v Si\c;lk Treatment Data 1181
Drcember 1985

lor higher initial population densities. This results mercial Bacillus thuringiensis. Can. Entomol. 116:
983-990.
in overestimation (trial 4) and underestimation Nord, J. C, G. L. DeBarr, L. R. Barber, J. C. Weath-
(trial 5) of the percent population reduction. erby, and IN. A. Overgaard. 1985. Low-volume
Hence, for unreplicated treatments made to re applications of azinphosmethyl, fenvalerate, and
duce population density, regardless of whether the permethrin for control of coneworms (Lepidoptera:
pretreatment population densities are actually Pyralidae) and seed bugs (Hemiptera: Coreidae and
equal in the check and treatment plots, designat Pentatomidae) in southern pine seed orchards. J.
ing the plot with the larger pretreatment density Econ. Entomol. 78. 445-450.
estimate for the check tends to produce conser Overhulser, D. L., G. E. Daterman, L. L. Sower, C.
vative estimates (underestimates) of population re Sartwell, and T. W. Koerber. 1980. Mating dis
ruption with synthetic sex attractants controls dam
duction. Using that plot for treatment tends to pro age by Eucosma sonomana (Lepidoptera: Tortrici
duce optimistic estimates (overestimates) of dae, Olethreutinae) in Pinus ponderosa plantations.
population reduction. For treatments made to in II. Aerially applied hollow fiber formulation. Can.
crease population density, conservative estimates Entomol. 112: 163-165.
(overestimates) of R tend to arise by designating Philogene, B. J. R., J. T. Arnason, C. W. Berg, F.
the plot with the smaller pretreatment density es Duval, D. Champagne, R. G. Taylor, L. C. Leitch,
timate as the check. This conclusion offers guid and P. Morand. 1985. Synthesis and evaluation of
ance for both experimental design and the inter the naturally occurring phototoxin, alphaterthienyl,
pretation of spray-trial data and other data. as a control agent for larvae of Aedes intrudens,
Aedes atropalpus (Diptera: Culicidae) and Simu-
lium verecundum (Diptera: Simuliidae). J. Econ.
Entomol. 78: 121-126.
Acknowledgment Pielou, E. C. 1969. An introduction to mathematical
We appreciate the constructive criticisms offered by ecology. Wiley, New York.
G. Grant, G. Howse, and B. Payandeh (Canadian For. Retnakaran, A. 1980. Effect of 3 new moult-inhib-
Serv.).
iting insect growth regulators on the spruce bud-
worm. J. Econ. Entomol. 73: 520-524.
1981. Toxicology and efficacy of insect growth reg
References Cited
ulators aerially applied against the spruce budworm
at Hearst (1978), Wawa (1979), and the French Riv
Abbott, W. S. 1925. A method of computing the ef er area (1980). Can. For. Serv. Inf. Rep. FPM-X-45.
fectiveness of an insecticide. J. Econ. Entomol. 18: Retnakaran, A., and G. G. Grant. 1985. Control of
265-267. the oak-leaf shredder Croesia semipurpurana (Kear-
Bostanian, N. J., A. Belanger, and I. Rivard. 1985. fott) (Lepidoptera: Tortricidae), by aerial application
Residues of four synthetic pyrethroids and azinphos- of diflubenzuron. Can. Entomol. 117: 363-369.
methyl on apple foliage and their toxicity to Ambly- Solomon, J. D. 1985. Comparative effectiveness of
seius fallacis (Acari: Phytoseiidae). Can. Entomol. gallery-injected insecticides and fumigants to control
117: 143-152. carpenterworms (Lepidoptera: Cossidae) and oak
Hurlbert, S. H. 1984. Pseudoreplication and the de clearwing borers (Lepidoptera: Sesiidae). J. Econ.
sign of ecological field experiments. Ecol. Monogr. Entomol. 78: 485-488.
54: 187-211. Sower, L. L., D. L. Overhulser, G. E. Daterman, C.
Krebs, C. J. 1972. Ecology: the experimental analysis Sartwell, D. E. Laws, and T. W. Koerber. 1982.
of distribution and abundance. Harper and Row, New Control of Eucosma sonomana by mating disruption
York. with svnthetic sex attractants. J. Econ. Entomol. 75:
Vloffitt, H. R., and P. H. Westigard. 1984. Suppres 315-318.
sion of the codling moth (Lepidoptera: Tortricidae) Youssef, N., and W. R. McManus. 1985. Captan: a
population on pears in southern Oregon through promising fungicide for management of chalkbrood
mating disruption with sex pheromone. J. Econ. diseases in the alfalfa leafcutting bee. J. Econ. Ento
Entomol. 77: 1513-1519. mol. 78: 428-431.
Morris, O. N. 1984. Field response of the spruce bud-
worm, Choristoneura fumiferana (Lepidoptera: Received for publication 4 February 1985; accepted
Tortricidae), to dosage and volume rates of com 26 August 1985.

View publication stats