- Email: [email protected]
Fumigation with carbonyl sulfide: a model for the interaction of concentration, time and temperature
Journal of Stored Products Research 37 (2001) 383–398
Fumigation with carbonyl sulfide: a model for the interaction of concentration, time and temperature G.L. Wellera,*, R. Mortonb a
Stored Grain Research Laboratory, CSIRO Entomology, GPO Box 1700, Canberra, ACT 2601, Australia b CSIRO Mathematical & Information Sciences, GPO Box 664, Canberra, ACT 2601, Australia Accepted 28 September 2000
Abstract The new fumigant carbonyl sulfide offers an alternative to both methyl bromide and phosphine as a grain fumigant. Separate mathematical models for levels of kill, based on quantitative toxicological studies were developed for adults and eggs of the rice weevil Sitophilus oryzae (L.). These models suggest that fumigation exposure times for carbonyl sulfide will be a compromise between those of methyl bromide (typically 24 h) and phosphine (7–10 d) to achieve a very high kill of all developmental stages. S. oryzae eggs were more difficult to kill with carbonyl sulfide fumigation than the adults. At 308C, a 25 g m3 fumigation killed 99.9% of adults in less than 1 d, but took 4 d to kill the same percentage of eggs. Models were generated to describe the mortality of adults at 10, 15, 20, 25 and 308C. From these models it is predicted that fumigation with carbonyl sulfide for 1–2 d at 30 g m3 will kill 99.9% of adults. Furthermore the models illustrate that fumigations with concentrations below 10 g m3 are unlikely to kill all adult S. oryzae. Significant variation was observed in the response of eggs to the fumigant over the temperature range of 10 to 308C. Models were generated to describe the mortality of eggs at 10, 15, 20, 25 and 308C. As the temperature was reduced below 258C, the time taken to achieve an effective fumigation increased. Extrapolating from the models, a 25 g m3 fumigation to control 99.9% of S. oryzae eggs will take 95 h (4 d) at 308C, 77 h (3.2 d) at 258C, 120 h (5 d) at 208C, 174 h (7.5 d) at 158C and about 290 h (11 d) at 108C. The role of temperature in the time taken to kill eggs with carbonyl sulfide cannot be ignored. In order to achieve the desired level of kill of all developmental stages, the fumigation rates need to be set according to the most difficult life stage to kill, in this instance, the egg stage. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: Carbonyl sulfide; Fumigation; Sitophilus oryzae (L); Mortality; Temperature
*Corresponding author. Fax: +61-2-62464202. E-mail address: [email protected] (G.L. Weller). 0022-474X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 2 - 4 7 4 X ( 0 0 ) 0 0 0 4 1 - 2
384
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
1. Introduction The use of toxic gases to disinfect grain and other stored commodities in storages is convenient and relatively inexpensive. Fumigation is often preferred to using grain protectants as a method of controlling insects in grain. Unlike the case with protectants, which require the commodity to be moved to apply the insecticide, a fumigant gas can be applied in situ, reaching all parts of the storage and stored commodity. The futures of the two major grain fumigants, methyl bromide and phosphine, are under threat. Methyl bromide, which is particularly valuable for rapid disinfestation (fumigation of grain with methyl bromide, including airing time, can be completed within 2 d) has been listed as an ozone depleting substance under the Montreal Protocol and will be phased out for most non-quarantine uses by the year 2005 (UNEP, 1996). The use of its major alternative phosphine is being threatened by the development of resistance in many insect pests of grain. At the recent 7th International Working Conference on Stored Product Protection in 1998 in Beijing, China, scientists from India, China, Vietnam, Poland and Australia reported on the occurrence and severity of phosphine resistance in their countries (Rajendren, 1999; Zeng, 1999; Bui, 1999; Ignatowicz, 1999; Bengston et al., 1999). Further threats to the use of phosphine include environmental and workspace restrictions, toxicity and effects on humans as well as accidents and misuse (Banks, 1994). If the use or effectiveness of phosphine were lost, grain fumigations would be dependent on carbon dioxide and low oxygen (or nitrogen) fumigation, both of which can take in excess of 14 d in a gas tight chamber to complete. The use of carbonyl sulfide (COS) as a fumigant was patented in 1993 (Banks et al., 1993) and described in detail by Desmarchelier (1994). Although COS is not a novel chemical, its use as a fumigant is new. As yet COS is not registered for use as a fumigant, but it is regarded as a possible alternative to methyl bromide. A number of studies have assessed the effectiveness of COS as a fumigant. Desmarchelier (1994), Plarre and Reichmuth (1997), Zettler et al. (1997, 1999) and Tan et al. (1999) have undertaken studies on the effectiveness of COS as a fumigant against a range of insects and mites. All concur that eggs are the most resistant life stage of coleopteran pests, followed by pupae and adults. Exposure times to achieve control of insect pests of grain are longer than those required for methyl bromide but may be faster than those required for other gaseous treatments. The pests found to be most tolerant to date are the Sitophilus species (S. granarius (L.) and S. oryzae (L.)) (Desmarchelier, 1994). In most fumigation scenarios, killing all life stages of infesting insects is important and doses are directed at the most tolerant stage. This infers that the focus of this experiment should be on the egg stage. However, in some situations, fumigations to kill only adults (and larvae) are undertaken. Such fumigations may be done as a ‘‘cosmetic’’ treatment or as a sequence of fumigations a couple of weeks apart, which allows for the development of the more tolerant life stages to a more susceptible stage. It is therefore appropriate to address doses required to control both adults and eggs separately. Previous studies have been almost entirely limited to a single temperature for each species; this study investigates the limitations of COS with respect to time, concentration and temperature against both eggs and adults of S. oryzae.
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
385
2. Materials and methods 2.1. Insects S. oryzae used during this study were sourced from a reference strain (LS2) held at the Stored Grain Research Laboratory, Canberra, Australia. Insects were cultured fortnightly at 258C on Australian Soft Wheat (var. Rosella) with a moisture content of 12% (wet basis). Adults collected for fumigation were 7–14 d post-emergence and eggs were 0–3.5 d old. Eggs were obtained by placing approximately 1200 adults (2.8 g) on 1.4 kg of wheat and incubating at 258C for 3 d prior to fumigation. After this time, adults were removed and the wheat mixed and divided using a Bo¨erner divider, to give 16 samples of 80–90 g. This procedure resulted in 300–500 emergent adults in non-fumigated exposures. 2.2. Fumigation Fumigations were carried out in glass desiccators (2.5–2.7 L) fitted with stainless-steel stirrers. The desiccator lids were sealed with glass stoppers fitted with a septum. The volume of each desiccator was determined from the mass of water it held. For adult exposures 100 g of the culture wheat (previously described) was placed in small crystallising dishes, which had been treated with poly-tetra-fluoro-ethylene (‘Fluon’) around the lip to restrict the movement of adult insects. These dishes were placed in the open desiccators and held at the fumigation temperature for 24 h prior to fumigation. Approximately 1 h prior to fumigation, approximately 100 adults (0.22 g) were added to the dishes and the desiccators were sealed. For egg exposures, the 80–90 g samples described above were placed in small uncovered culture jars, and in turn placed in the desiccators. Both the infested wheat and desiccators were held at the fumigation temperature overnight prior to fumigation. Fumigations were undertaken at a minimum of 4 concentrations over 5 time periods at each temperature. The time/concentration combinations used were chosen to reflect a continuous range of mortality from 0–100% and were therefore dependent on the temperature and life stage being tested. Each desiccator was dosed by removing a measured quantity of air and then introducing the same volume of gaseous COS through the septa with a gas tight syringe. The atmosphere in the desiccator was mixed using the magnetic stirrer for 5 min. COS was sourced from a compressed gas cylinder (BOC Gases Australia Ltd., Chatswood, NSW). Gas was applied in the range of 5–25 g m3 for adult exposures and 5–40 g m3 for egg exposures. On completion of fumigation the desiccators and contents were aired in a fume hood for approximately 1 h. After this time the insects and grain were transferred to culture jars and placed at 258C, 60% r.h. for the recovery period. Adult mortality was initially assessed at 24 h, 3, 5 and 7 d post fumigation. There was no evidence of narcosis at the doses applied. The number of dead adults rose from the 24 h count to the 3 d count and remained static after that time. For convenience all mortality assessments were undertaken at 7 d. Eggs were incubated at 258C, 60% r.h. for 9 weeks post-fumigation. Adult emergence was recorded at 7, 8 and 9 weeks post-fumigation. There was no evidence of delayed emergence from fumigated eggs, relative to the controls for each temperature.
386
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
The survival of eggs was calculated by comparing the numbers of adults emerging from treated replicates up to 9 weeks after the exposure, with those emerging from untreated (control) replicates. 2.3. Gas analysis Purity of the COS supplied by BOC was determined using a Gow Mac (Model 11–625) gas density balance on a Tracor (MT150) gas chromatograph fitted with a Porapak Q 80/100 column. COS was found to be 96.0% v/v with air and carbon dioxide making up the remainder. COS concentrations during fumigations were measured against standards using a Tracor (MT-220) GC fitted with a flame photometric detector (FPD)(sulphur mode) and a HayeSep Q 80/100 mesh column run at 1108C. The concentration of COS was measured within 30 min of the start of fumigation and again within 30 min prior to the end of the fumigation. Oxygen and carbon dioxide concentrations in the fumigation chambers were monitored daily during the fumigation. Gas samples (0.3 ml) were taken from the desiccators via the septa after stirring for approximately 5 min. These were analysed on a Fisher model 1200 Gas Partitioner fitted with an 80–100 mesh Column packTM PQ (6.5 ft 1/8 in.) and a 60–80 mesh molecular sieve 13X (11 ft 3/16 in.) column in series, and a thermal conductivity detector. The oven and detector were both run at 508C and helium at 30 ml min1 was used as the carrier gas. Concentrations were calculated from the peak areas using a Hewlett Packard Reporting Integrator model 3390A calibrated against standard gas mixtures. 2.4. Statistical methods For adults, a logistic regression was performed. The probability of death p was expressed as a linear model on the logit scale, logit ðpÞ ¼ lnfp=ð1 pÞg. Preliminary modelling indicated that the logit scale was better than the probit scale. It was assumed that the number Y of insects killed had expectation EðYÞ ¼ np and variance varðYÞ ¼ fnpð1 pÞ, where n is the number of insects exposed and f is a variance scale factor to allow for extra-binomial variation (i.e. f > 1 indicates that the variance is greater than that attributable to the binomial distribution). Wadley’s problem (Finney, 1980, Chap. 10) was encountered with the egg data. That is, n is not known but must be estimated from untreated control samples. It was reasonable to assume that the relevant control exposures would have the same expected numbers. However at the longer exposures (7–10 d) at 108C, the emergence was greatly reduced compared with the shorter exposures (1–4 d). Such ‘‘cold death’’ is not unexpected, as Fields (1992) reports that below 138C most stored product insects are unable to complete their development and produce offspring and eventually die. The 108C data were deleted from the analysis as they do not represent the same biological situation as do the higher temperature data. If Y is the number of survivors; then EðYÞ ¼ nð1 pÞ and varðYÞ ¼ fnð1 pÞ, where n is the expected number of survivors for the controls, p is the mortality expressed as a linear model on the logit scale and f is again a variance scale factor to allow for extra-Poisson variation. As with the adults, the logit was preferred to the probit scale. Estimation of the regression parameters is by maximum quasi-likelihood, and f is estimated by the mean residual deviance. Hypothesis tests use the deviance ratios, which have approximately an
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
387
F-distribution; approximate standard errors for the regression parameter estimates are calculated from the usual quasi-likelihood theory (McCullagh and Nelder, 1989). Computing was performed using Genstat 5 (Genstat 5 Committee, 1993).
3. Results and modelling 3.1. Gas concentrations Initial gas concentrations were measured approximately 30 min after dosing. The observed concentration was found to be within 5% of the targeted concentration, generally exceeding the target concentration due to the gas addition being calculated on an empty desiccator. No loss due to sorption was observed in the 10 and 158C fumigations; at the higher temperatures the head space concentration fell slightly to approximately 90% of the initial concentration over 7 days. For the purposes of modelling, the average concentration for each exposure was used. 3.2. Adults The survival of S. oryzae adults exposed to COS varied with concentration, time and temperature. Levels of carbon dioxide in the fumigation chambers rose over the duration of the exposure. In chambers where the level of kill was greater than 40%, the level of carbon dioxide did not exceeded 2%. Where mortality was low, fumigations were terminated when the concentration of carbon dioxide reached 4%. This limited the time of exposure for low doses of COS (less than 10 g m3). The survival for each concentration/time combination tested at 10, 15, 20, 25 and 308C is shown in Fig. 1. At each temperature, progressively lower concentrations took longer to achieve the same level of mortality, suggesting that an interaction between concentration and time is determining the level of mortality. The increasing survival at lower concentrations is suggestive of a threshold concentration, below which there is no effect on survival. The simplest model for the action of fumigant gases is that the product of the concentration (C) and the exposure time ðtÞ produces a constant level, i.e. Ct ¼ k of kill k (Haber, 1924), which has become known as Haber’s Rule. There has been considerable debate about the usefulness of Haber’s Rule and for what gases and under what conditions it applies. However, given small ranges for each of concentration, time and temperature, Haber’s rule could apply for most fumigant gases. While the simplicity of such a model is tempting, it could severely limit the range in which a fumigant is deemed useful. A model for mortality will identify the full range of Ct combinations at which this fumigant can be used and identify the limitations and conditions where fumigation may fail to kill all insects and/or select for tolerant insects. To test the adequacy of Haber’s rule as a model for these data, the term lnðCtÞ was fitted. This basic model was improved by the addition of subsequent terms. Adding the term lnðtÞ substantially reduced the residual deviance; however, this function was not linear and the term was better represented by lnðt þ 9:12Þ. The constant 9.12 was determined by maximum
388
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
Fig. 1. Survival of adult Sitophilus oryzae fumigated with carbonyl sulfide at a range of fixed concentrations and times at 10, 15, 20, 25 and 308C.
quasi-likelihood when this term and its interaction were included in the model. This feature of the model has no apparent mechanistic or biological explanation. The terms in order of fit for the adult mortality model and the analyses of deviance are given in Table 1. The analyses of deviance tables give the relative magnitude of the improved fit due to
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
389
adding each new term to the model. Tests of significance of each term in the models are based on t-statistics; these show the significance of each term allowing for all the others. The analysis of deviance table gives the significance of the fitted term allowing for the previous terms fitted but ignoring the subsequent ones. This leaves the general model lnfp=ð1 pÞg ¼ a þ blnðCtÞ þ clnðt þ 9:12Þ;
ð1Þ
where p in the probability of death, a, b, and c are constants, C is concentration in g m3 and t is time in h. The estimation of the value of Ct which achieves p 100% mortality is lnðCtÞ ¼ ½ln p=ð1 pÞ a clnðt þ 9:12Þ =b:
ð2Þ
When p ¼ 0:95, 0.99 and 0.999, lnðp=ð1 pÞÞ ¼ 2:944, 4.595 and 6.907, respectively. Temperature effects are statistically significant but might be considered small for practical purposes. If temperature effects are ignored and all data pooled, the fitted model for mortality is lnfp=ð1 pÞg ¼ 18:864 þ 6:296 lnðCtÞ 3:815 lnðt þ 9:12Þ:
ð3Þ
This model is illustrated in Fig. 2. A theoretical curve for 99% kill (which is within the domain of the data) and an extrapolated curve for 99.9% kill are plotted against the experimental data. Comparing this combined model with those for individual temperatures (Fig. 3), it is clear that the general model is not a good representation of mortality observed at extreme values such as 158C. However, under most conditions, a concentration in excess of 15 g m3 for 48 h or longer will kill at least 99% of adult S. oryzae. To increase the level of kill to 99.9%, given the same concentration of 15 g m3, the predicted duration of fumigation is more than 96 h (4 d). If the duration of fumigation were required to be less than 48 h, the dose required for 99.9% kill would need to be increased to more than 22 g m3. Caution should be exercised in interpreting from the 99.9% model as this is an extrapolation from the model and goes beyond the domain of the data. The tailing of the model at each axis suggests that concentrations required to kill adults in less than one day are likely to be very high and that concentrations of the fumigant below 10 g m3 are unlikely to achieve a 99.9% kill of adults. Table 1 Analysis of deviance for terms used to create the model for adult mortality following carbonyl sulfide fumigation. C is the concentration of carbonyl sulfide in g m3, t is time in hours and T is the temperature in 8C Terms in order of fit
d.f.
Deviance
Mean deviance
Deviance ratio Fa
lnðCtÞ lnðt þ 9:12Þ T20 lnðCtÞ ðT 20Þ lnðtþ9:12Þ ðT 20Þ Residual Total
1 1 1 1 1 176 181
5778 3280 65 269 228 793 10,413
5778 3280 65 269 228 4.51 58.53
1282 727 14 60 51
Adjusted R2 = 92.3%. a All t-values have P50:001 except for t=1.21 for which P=2.226.
390
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
Fig. 2. Theoretical curves for 99 and 99.9% kill, generated from the mathematical model for adult mortality for the temperature range 10–308C, plotted against observations of 100% (*), 95–99.9% (&), and 85–94.9% ( ) kill. For comparison, a plot of Ct ¼ 1284 g h m3 is included.
Fig. 3. Theoretical curves for 99% kill of adult Sitophilus oryzae fumigated with carbonyl sulfide at 10, 15, 20, 25 and 308C.
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
391
For comparison, a plot for best fit of the data to a Ct ¼ k model for 99.9% kill is included in Fig. 3 (Ct¼ 1284 g h m3 ). Comparison of the model to this plot shows that there is considerable variation between the generated model and a straight Ct model. If a Ct model were applied, fumigations undertaken for longer than 120 h periods would be at risk of underdosing and thereby increase the likelihood of survival and the ultimate development of COS-resistant insects. If temperature ðTÞ is included in the model, the major contributory components are the linear effect of temperature and the linear component of its interaction with the terms lnðCtÞ and lnðt þ 9:12Þ. The fitted model, with 208C as the reference temperature, then becomes lnfp=ð1 pÞg ¼ 20:644 þ 6:296lnðCtÞ 3:887lnðt þ 9:12Þ þ 0:5974 ð4Þ ðT 20Þ þ 0:0316ðT 20ÞlnðCtÞ 0:1938ðT 20Þlnðt þ 9:12Þ Theoretical curves for 99% kill at each of the tested temperatures are plotted as concentration against time in Fig. 3. These plots show how the relationship between concentration and time change with temperature. It appears that higher doses require a shorter exposure period to kill the same percentage at higher temperatures. According to this model, a fumigation of 25 g m3 will take about 12 h to kill 99% at 308C and progressively longer as temperatures decrease, taking more than 24 h to achieve the same level of kill at 108C. This can be interpreted from the raw data by a comparison of the level of kill achieved at one concentration and one time across the temperature range. For example, a 24-h fumigation of 15 g m3 killed 99% at 308C, 98% at 258C, 97% at 208C, 95% at 158C and only 92% at 108C. Perhaps of more importance is the difference between the minimal dose to achieve 99% kill over longer fumigations. Using 96 h as an example, it is clear that the concentration required to kill 99% is much higher at 308C than at 108C. Details of the parameter estimates for fitting the main terms for the adult models are given in Table 2. 3.3. Eggs As with the adults, egg survival varied with concentration, time and temperature. The level of carbon dioxide in the fumigation chambers never exceeded 0.5%. The survival for each Ct combination at 10, 15, 20, 25 and 308C is shown in Fig. 4. Table 2 Parameter estimates for the model for adult Sitophilus oryzae mortality following carbonyl sulfide fumigation. C is the concentration of carbonyl sulfide in g m3, t is time in hours and T is the temperature in 8C Term
Estimate
SE
T176a
Constant lnðCtÞ lnðt þ 9:12Þ T20 lnðCtÞ ðT 20Þ lnðt þ 9:12Þ ðT20Þ
20.644 6.296 3.887 0.5974 0.0316 0.1938
0.779 0.224 0.178 0.0956 0.0260 0.0268
26.49 28.12 21.83 6.25 1.21 7.24
a All t-values have P50:001 except for t ¼ 1:21 for which P ¼ 0:226.
392
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
In a similar vein to the modelling that was undertaken on the adult data, Ct was assumed to be the major term and to test whether C and t had equal impact the terms lnðCtÞ and lnðtÞ were used. Arbitrary smooth curves (splines) were fitted to see if there was any residual curvature; the splines
Fig. 4. Survival of Sitophilus oryzae eggs fumigated with carbonyl sulfide at a range of fixed concentrations and times at 10, 15, 20, 25 and 308C.
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
393
were almost straight, which indicated that log terms were suitable for use in modelling the data. Slight curvature in lnðtÞ was found which was reduced by adding a constant, for consistency the same constant of 9.12 (from the adult model) was used. Thus the term became lnðt þ 9:12Þ. Models for eggs that assume that each temperature level is different were analysed in Genstat using the Genstat procedure ‘probitanalysis’. To fit a model to all temperatures from 15 to 308C a novel Genstat code ‘fitnonlinear’ was used. The analyses of deviance for the main terms of egg mortality model are given in Table 3. Eq. (1) was again used as the basic logit model for mortality. Temperature, when fitted, was treated as a factor, because the linear term T 20 did not produce an adequate fit. Table 3 Analysis of deviance for terms used to create the model for egg mortality following carbonyl sulfide fumigation. C is the concentration of carbonyl sulfide in g m3, t is time in hours and T is temperature in 8C Terms in order of fit
d.f.
Deviance
Mean deviance
Deviance ratio F a
ln(Ct) ln(t þ 9:12) T lnðCtÞT lnðt þ 9:12ÞT Residual Total
1 1 3 3 3 88 99
4868 256 847 34 197 343 6544
4868 256 282 11 66 3.90 66.10
1248 66 72 2.90 16.92
a
All F-values have P50:001 except for F ¼ 2:90 for which P ¼ 0:039.
Table 4 Estimated parameters ( SE) for constants (a, b and c) in the model ‘‘logitðpÞ ¼ a þ blnðCtÞ þ clnðt þ 9:12Þ’’ which describes the mortality of Sitophilus oryzae eggs fumigated with carbonyl sulfide. Where p is the probability of death, C is the concentration of carbonyl sulfide in g m3, t is time in hours, n is estimated control emergence, and j is the variance scale parameter. The last row fits common values of a, b and c for all temperatures in the range 15–308C Temp.8C
n
a
b
c
j
10
562.1 ( 26.6) 368.2 ( 12.7) 286.8 ( 8.18) 126.8 ( 5.31) 351.9 ( 17.2) Various
29.214 ( 3.434) 43.000 ( 6.183) 36.041 ( 3.473) 37.469 ( 4.657) 29.373 ( 3.542) 23.665 ( 2.384)
2.703 ( 0.330) 3.538 ( 0.568) 4.454 ( 0.495) 5.189 ( 0.659) 4.778 ( 0.662) 2.941 ( 0.355)
2.121 ( 0.304) 3.889 ( 0.650) 1.499 ( 0.270) 1.160 ( 0.265) 0.180 ( 0.367) 1.037 ( 0.263)
10.62
15 20 25 30 15–30
5.413 2.321 1.773 6.535 14.62
394
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
Fig. 5. Theoretical curves for 99% kill of Sitophilus oryzae eggs generated from the mathematical models for egg mortality. A separate curve is given for each temperature 10, 15, 20, 25, 308C, and the combined temperature range 15–308C.
Table 5 Time to kill 99.9% of Sitophilus oryzae eggs (calculated from the individual temperature models), when fumigated at a rate of 25 g m3 carbonyl sulfide, at 10, 15, 20, 25 and 308C Temperature (8C)
Concentration (g m3)
Time (h)
Approximate time (d)
10 15 20 25 30
25 25 25 25 25
290 174 120 77 95
11 7.5 5.0 3.2 4.0
Details of the parameter estimates for fitting the main terms for the egg model for each temperature are given in Table 4. Separate models for 99% kill fitted for each temperature are compared with that fitted for all of the data from 15 to 308C in Fig. 5. From these theoretical curves it appears that COS is most effective in the range of 20–308C. Fumigation at 15 or 108C takes longer to achieve the same level of kill given the same concentration. The longer duration could be due to a reduction in metabolism or slowing of development at lower temperatures. Not surprisingly the model for the combined temperatures (15–308C) most closely mimics the 158C curve. The requirement for longer fumigations at 158C pushes the time required to achieve 99% kill out to more than 96 h (4 d) at 25 g m3, as opposed to about 48 h for 25 g m3 at 258C. As was the case with the adult model, the combined temperature model is not a good representation of mortality observed at
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
395
Fig. 6. Theoretical curves for 99 and 99.9% kill, generated from the mathematical model for egg mortality for the temperature range 15–308C, plotted against observations of 100% (*), 95–99.9% (&), and 85–94.9% ( ) kill. For comparison, a plot of Ct ¼ 4512 g h m3 is included.
extreme values such as low concentrations and low temperature (158C). For instance, if a dose rate of 15 g m3 was chosen for the fumigation of S. oryzae eggs from the general model, and applied at 158C, there could be strong selection for tolerant insects and a likelihood of emergent resistance to COS. It is therefore preferable to apply COS according to the individual temperature models rather than relying on the combined temperature model. Extrapolating from the individual temperature models, Table 5 gives the theoretical times taken to kill 99.9% of S. oryzae eggs with a concentration of 25 g m3 at 10, 15, 20, 25 and 308C. An extrapolated plot for 99.9% kill from the combined temperature model is compared with the experimental data in Fig. 6. Although 99.9% kill is beyond the domain of the data collected in this experiment it gives an idea of how much longer a fumigation will need to be to achieve the increased level of kill. For comparison, a plot for best fit of the data to a Ct ¼ k model for 99.9% kill is included in Fig. 6 (Ct¼ 4512 g h m3 ). In this instance a basic Ct model would underestimate the dose required to achieve 99.9% kill up until 264 h after which the dose would be overestimated. 4. Discussion From the observed mortality of S. oryzae eggs and adults, fumigated with COS, mathematical models have been developed which can be used to predict the concentrations and times
396
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
required for various levels of kill over a range of temperatures. Now the debate can focus on what level of kill is acceptable and safe in terms of avoiding the possible development of resistance. Fumigations to kill only adults (and larvae) are not recommended for the long-term control of insect pests, but may be applied under certain conditions, or as a sequence of fumigations a couple of weeks apart, allowing for the development of the more tolerant life stages to a more susceptible stage. The models developed in this paper for the control of S. oryzae adults indicate that within the temperature range of 15–308C, a 24 h exposure at 30 g m3 will kill 99% and that concentrations below 10 g m3 are unlikely to result in complete control of S. oryzae even with extended exposures such as 10 to 14 days. Fumigations to control all life stages of pests are routinely undertaken and setting an appropriate concentration and time regime for the fumigation is vital. S. oryzae eggs were more difficult to kill with carbonyl sulfide fumigation than the adults. At 308C, a 25 g m3 fumigation is predicted to kill 99.9% of adults in less than 1 d, but will theoretically take 4 d to kill the same percentage of eggs. The models developed for the control of S. oryzae eggs at each temperature emphasise the importance of understanding how environmental factors such as temperature can affect the outcome of a given fumigation. Something as simple as 58C temperature change can alter a proposed 5 d fumigation at 208C (25 g m3 to achieve 99.9% kill) to a 7.5 d fumigation at 158C (to achieve the same level of kill from the same concentration). That is another 2.5 d to maintain the concentration and withhold the goods. Temperature is only one of the considerations that can affect the outcome of fumigations. Many other factors have not been considered, including sorption by the commodity and other target species. Although it was assumed that S. oryzae eggs are the most tolerant life stage of one of the more tolerant species of stored product insects (Desmarchelier, 1994; Tan et al., 1999; Plarre and Reichmuth, 1997; Zettler et al., 1997; Weller, unpublished data), it is possible that a species not yet tested against COS will be more tolerant. With this in mind, it should not be assumed that the models generated for the use of COS in this paper will be applicable to all insects. In summary, COS has potential as an alternative to both methyl bromide and phosphine for fumigation of stored products. Although the Ct product required for a successful fumigation with COS is much higher than is typical for methyl bromide (approximately 2000 g h m3 for COS compared with 200 g h m3 for methyl bromide), the higher Ct product is achievable in practice because of lower sorption of COS than methyl bromide onto structures and commodities (Desmarchelier, 1994; Ren, 1996). Unfortunately, fumigations undertaken with COS will be significantly longer that those with methyl bromide but will be shorter to achieve the same level of kill than with phosphine, carbon dioxide or low oxygen (nitrogen).
Acknowledgements The research reported in this paper was supported by funds from the Partners to the Stored Grain Research Laboratory Agreement and CSIRO Australia. The authors are also grateful to Jim Desmarchelier, Peter Annis, Jonathan Banks, Albert Trajstman and Yong Lin Ren for their critical review and comments of this manuscript.
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
397
References Banks, H.J., 1994. Fumigation } an endangered technology? In: Highley, E., Wright, E.J., Banks, H.J., Champ, B.R. (Eds.), Stored Product Protection, Proceedings of the 6th International Working Conference on Stored-Product Protection, Canberra, Australia, 17–23 April 1994. CAB International, Wallingford, Oxon, UK, pp. 2–6. Banks, H.J., Desmarchelier, J.M., Ren, Y.L., 1993. Carbonyl sulfide fumigant and method of fumigation. International Application Published under the Patent Cooperation Treaty (PCT), WO93/13659. Bengston, M., Acada, M., Daglish, G.D., Collins P.J., 1999. Phosphine resistance in the Asia/Pacific region. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 631–634. Bui Cong Hien, 1999. Some initial results on phosphine resistance of major product insect pests in Vietnam. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 648–652. Desmarchelier, J.M., 1994. Carbonyl sulfide as a fumigant for control of insects and mites. In: Highley, E., Wright, E.J., Banks, H.J., Champ, B.R. (Eds), Stored Product Protection, Proceedings of the 6th International Working Conference on Stored-Product Protection, Canberra, Australia, 17–23 April 1994. CAB International, Wallingford, Oxon, UK, pp.78–82. Fields, P.G., 1992. The control of stored-product insects and mites with extreme temperatures. Journal of Stored Products Research 28, 89–118. Finney, D.J., 1980. Probit Analysis, 3rd ed. Cambridge University Press, Cambridge. Genstat 5 Committee, 1993. Genstat 5 Release 3 Reference Manual. Clarendon Press, Oxford. Haber, F., 1924. Fu¨nf Vortra¨ge aus den Jahren 1920–1923. Springer, Berlin, p. 98S. Ignatowicz, S., 1999. Cases of phosphine resistance for grain weevil, Sitophilus granarius found in Poland. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 625–630. McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models, 2nd ed. Chapman & Hall, London. Plarre, R., Reichmuth, C., 1997. Effects of carbonyl sulfide on Sitophilus granarius (L.) (Coleoptera: Curculionidae), Fusarium culmorum and Fusarium avenaceum (SACC.) (Deuteromycotina: Hyphomycetes), and corrosion on copper. In: Donahaye, E.J., Navarro, S., Varnava, A. (Eds.), Proceedings of the International Conference on Controlled Atmosphere and Fumigation in Stored products, Nicosia, Cyprus, 21–26 April 1996. Printco Ltd., Nicosia, Cyprus, pp. 59–71. Rajendran, S., 1999. Phosphine resistance in stored grain insect pests in India. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 635–641. Ren, Y.L., 1996. Carbonyl sulfide as a fumigant for grain and timber. Ph.D. Thesis, University of Canberra, Australia. Tan Xianchang, Hou Xingwei, Cheng Lizheng, Wu Jianchun, 1999. Research on carbonyl sulfide as a fumigant for control of stored grain insects. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 567–571. U.N.E.P., 1996. Eighth Meeting of the Parties to the Montreal Protocol on Substances that Deplete the Ozone Layer. Vienna, November 1996. Zeng, L., 1999. Development and counter measures of phosphine resistance in stored grain insects in Guangdong of China. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product
398
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 642–647. Zettler, J.L., Leesch, J.G., Gill, R.F., Mackey, B.E., 1997. Toxicity of carbonyl sulfide to stored product insects. Journal of Economic Entomology 90, 832–836. Zettler, J.L., Leesch, J.G., Gill, R.F., Tebbets, J.G., 1999. Chemical alternatives for methyl bromide and phosphine treatments for dried fruits and nuts In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 554–561.
Fumigation with carbonyl sulfide: a model for the interaction of concentration, time and temperature G.L. Wellera,*, R. Mortonb a
Stored Grain Research Laboratory, CSIRO Entomology, GPO Box 1700, Canberra, ACT 2601, Australia b CSIRO Mathematical & Information Sciences, GPO Box 664, Canberra, ACT 2601, Australia Accepted 28 September 2000
Abstract The new fumigant carbonyl sulfide offers an alternative to both methyl bromide and phosphine as a grain fumigant. Separate mathematical models for levels of kill, based on quantitative toxicological studies were developed for adults and eggs of the rice weevil Sitophilus oryzae (L.). These models suggest that fumigation exposure times for carbonyl sulfide will be a compromise between those of methyl bromide (typically 24 h) and phosphine (7–10 d) to achieve a very high kill of all developmental stages. S. oryzae eggs were more difficult to kill with carbonyl sulfide fumigation than the adults. At 308C, a 25 g m3 fumigation killed 99.9% of adults in less than 1 d, but took 4 d to kill the same percentage of eggs. Models were generated to describe the mortality of adults at 10, 15, 20, 25 and 308C. From these models it is predicted that fumigation with carbonyl sulfide for 1–2 d at 30 g m3 will kill 99.9% of adults. Furthermore the models illustrate that fumigations with concentrations below 10 g m3 are unlikely to kill all adult S. oryzae. Significant variation was observed in the response of eggs to the fumigant over the temperature range of 10 to 308C. Models were generated to describe the mortality of eggs at 10, 15, 20, 25 and 308C. As the temperature was reduced below 258C, the time taken to achieve an effective fumigation increased. Extrapolating from the models, a 25 g m3 fumigation to control 99.9% of S. oryzae eggs will take 95 h (4 d) at 308C, 77 h (3.2 d) at 258C, 120 h (5 d) at 208C, 174 h (7.5 d) at 158C and about 290 h (11 d) at 108C. The role of temperature in the time taken to kill eggs with carbonyl sulfide cannot be ignored. In order to achieve the desired level of kill of all developmental stages, the fumigation rates need to be set according to the most difficult life stage to kill, in this instance, the egg stage. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: Carbonyl sulfide; Fumigation; Sitophilus oryzae (L); Mortality; Temperature
*Corresponding author. Fax: +61-2-62464202. E-mail address: [email protected] (G.L. Weller). 0022-474X/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 2 - 4 7 4 X ( 0 0 ) 0 0 0 4 1 - 2
384
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
1. Introduction The use of toxic gases to disinfect grain and other stored commodities in storages is convenient and relatively inexpensive. Fumigation is often preferred to using grain protectants as a method of controlling insects in grain. Unlike the case with protectants, which require the commodity to be moved to apply the insecticide, a fumigant gas can be applied in situ, reaching all parts of the storage and stored commodity. The futures of the two major grain fumigants, methyl bromide and phosphine, are under threat. Methyl bromide, which is particularly valuable for rapid disinfestation (fumigation of grain with methyl bromide, including airing time, can be completed within 2 d) has been listed as an ozone depleting substance under the Montreal Protocol and will be phased out for most non-quarantine uses by the year 2005 (UNEP, 1996). The use of its major alternative phosphine is being threatened by the development of resistance in many insect pests of grain. At the recent 7th International Working Conference on Stored Product Protection in 1998 in Beijing, China, scientists from India, China, Vietnam, Poland and Australia reported on the occurrence and severity of phosphine resistance in their countries (Rajendren, 1999; Zeng, 1999; Bui, 1999; Ignatowicz, 1999; Bengston et al., 1999). Further threats to the use of phosphine include environmental and workspace restrictions, toxicity and effects on humans as well as accidents and misuse (Banks, 1994). If the use or effectiveness of phosphine were lost, grain fumigations would be dependent on carbon dioxide and low oxygen (or nitrogen) fumigation, both of which can take in excess of 14 d in a gas tight chamber to complete. The use of carbonyl sulfide (COS) as a fumigant was patented in 1993 (Banks et al., 1993) and described in detail by Desmarchelier (1994). Although COS is not a novel chemical, its use as a fumigant is new. As yet COS is not registered for use as a fumigant, but it is regarded as a possible alternative to methyl bromide. A number of studies have assessed the effectiveness of COS as a fumigant. Desmarchelier (1994), Plarre and Reichmuth (1997), Zettler et al. (1997, 1999) and Tan et al. (1999) have undertaken studies on the effectiveness of COS as a fumigant against a range of insects and mites. All concur that eggs are the most resistant life stage of coleopteran pests, followed by pupae and adults. Exposure times to achieve control of insect pests of grain are longer than those required for methyl bromide but may be faster than those required for other gaseous treatments. The pests found to be most tolerant to date are the Sitophilus species (S. granarius (L.) and S. oryzae (L.)) (Desmarchelier, 1994). In most fumigation scenarios, killing all life stages of infesting insects is important and doses are directed at the most tolerant stage. This infers that the focus of this experiment should be on the egg stage. However, in some situations, fumigations to kill only adults (and larvae) are undertaken. Such fumigations may be done as a ‘‘cosmetic’’ treatment or as a sequence of fumigations a couple of weeks apart, which allows for the development of the more tolerant life stages to a more susceptible stage. It is therefore appropriate to address doses required to control both adults and eggs separately. Previous studies have been almost entirely limited to a single temperature for each species; this study investigates the limitations of COS with respect to time, concentration and temperature against both eggs and adults of S. oryzae.
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
385
2. Materials and methods 2.1. Insects S. oryzae used during this study were sourced from a reference strain (LS2) held at the Stored Grain Research Laboratory, Canberra, Australia. Insects were cultured fortnightly at 258C on Australian Soft Wheat (var. Rosella) with a moisture content of 12% (wet basis). Adults collected for fumigation were 7–14 d post-emergence and eggs were 0–3.5 d old. Eggs were obtained by placing approximately 1200 adults (2.8 g) on 1.4 kg of wheat and incubating at 258C for 3 d prior to fumigation. After this time, adults were removed and the wheat mixed and divided using a Bo¨erner divider, to give 16 samples of 80–90 g. This procedure resulted in 300–500 emergent adults in non-fumigated exposures. 2.2. Fumigation Fumigations were carried out in glass desiccators (2.5–2.7 L) fitted with stainless-steel stirrers. The desiccator lids were sealed with glass stoppers fitted with a septum. The volume of each desiccator was determined from the mass of water it held. For adult exposures 100 g of the culture wheat (previously described) was placed in small crystallising dishes, which had been treated with poly-tetra-fluoro-ethylene (‘Fluon’) around the lip to restrict the movement of adult insects. These dishes were placed in the open desiccators and held at the fumigation temperature for 24 h prior to fumigation. Approximately 1 h prior to fumigation, approximately 100 adults (0.22 g) were added to the dishes and the desiccators were sealed. For egg exposures, the 80–90 g samples described above were placed in small uncovered culture jars, and in turn placed in the desiccators. Both the infested wheat and desiccators were held at the fumigation temperature overnight prior to fumigation. Fumigations were undertaken at a minimum of 4 concentrations over 5 time periods at each temperature. The time/concentration combinations used were chosen to reflect a continuous range of mortality from 0–100% and were therefore dependent on the temperature and life stage being tested. Each desiccator was dosed by removing a measured quantity of air and then introducing the same volume of gaseous COS through the septa with a gas tight syringe. The atmosphere in the desiccator was mixed using the magnetic stirrer for 5 min. COS was sourced from a compressed gas cylinder (BOC Gases Australia Ltd., Chatswood, NSW). Gas was applied in the range of 5–25 g m3 for adult exposures and 5–40 g m3 for egg exposures. On completion of fumigation the desiccators and contents were aired in a fume hood for approximately 1 h. After this time the insects and grain were transferred to culture jars and placed at 258C, 60% r.h. for the recovery period. Adult mortality was initially assessed at 24 h, 3, 5 and 7 d post fumigation. There was no evidence of narcosis at the doses applied. The number of dead adults rose from the 24 h count to the 3 d count and remained static after that time. For convenience all mortality assessments were undertaken at 7 d. Eggs were incubated at 258C, 60% r.h. for 9 weeks post-fumigation. Adult emergence was recorded at 7, 8 and 9 weeks post-fumigation. There was no evidence of delayed emergence from fumigated eggs, relative to the controls for each temperature.
386
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
The survival of eggs was calculated by comparing the numbers of adults emerging from treated replicates up to 9 weeks after the exposure, with those emerging from untreated (control) replicates. 2.3. Gas analysis Purity of the COS supplied by BOC was determined using a Gow Mac (Model 11–625) gas density balance on a Tracor (MT150) gas chromatograph fitted with a Porapak Q 80/100 column. COS was found to be 96.0% v/v with air and carbon dioxide making up the remainder. COS concentrations during fumigations were measured against standards using a Tracor (MT-220) GC fitted with a flame photometric detector (FPD)(sulphur mode) and a HayeSep Q 80/100 mesh column run at 1108C. The concentration of COS was measured within 30 min of the start of fumigation and again within 30 min prior to the end of the fumigation. Oxygen and carbon dioxide concentrations in the fumigation chambers were monitored daily during the fumigation. Gas samples (0.3 ml) were taken from the desiccators via the septa after stirring for approximately 5 min. These were analysed on a Fisher model 1200 Gas Partitioner fitted with an 80–100 mesh Column packTM PQ (6.5 ft 1/8 in.) and a 60–80 mesh molecular sieve 13X (11 ft 3/16 in.) column in series, and a thermal conductivity detector. The oven and detector were both run at 508C and helium at 30 ml min1 was used as the carrier gas. Concentrations were calculated from the peak areas using a Hewlett Packard Reporting Integrator model 3390A calibrated against standard gas mixtures. 2.4. Statistical methods For adults, a logistic regression was performed. The probability of death p was expressed as a linear model on the logit scale, logit ðpÞ ¼ lnfp=ð1 pÞg. Preliminary modelling indicated that the logit scale was better than the probit scale. It was assumed that the number Y of insects killed had expectation EðYÞ ¼ np and variance varðYÞ ¼ fnpð1 pÞ, where n is the number of insects exposed and f is a variance scale factor to allow for extra-binomial variation (i.e. f > 1 indicates that the variance is greater than that attributable to the binomial distribution). Wadley’s problem (Finney, 1980, Chap. 10) was encountered with the egg data. That is, n is not known but must be estimated from untreated control samples. It was reasonable to assume that the relevant control exposures would have the same expected numbers. However at the longer exposures (7–10 d) at 108C, the emergence was greatly reduced compared with the shorter exposures (1–4 d). Such ‘‘cold death’’ is not unexpected, as Fields (1992) reports that below 138C most stored product insects are unable to complete their development and produce offspring and eventually die. The 108C data were deleted from the analysis as they do not represent the same biological situation as do the higher temperature data. If Y is the number of survivors; then EðYÞ ¼ nð1 pÞ and varðYÞ ¼ fnð1 pÞ, where n is the expected number of survivors for the controls, p is the mortality expressed as a linear model on the logit scale and f is again a variance scale factor to allow for extra-Poisson variation. As with the adults, the logit was preferred to the probit scale. Estimation of the regression parameters is by maximum quasi-likelihood, and f is estimated by the mean residual deviance. Hypothesis tests use the deviance ratios, which have approximately an
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
387
F-distribution; approximate standard errors for the regression parameter estimates are calculated from the usual quasi-likelihood theory (McCullagh and Nelder, 1989). Computing was performed using Genstat 5 (Genstat 5 Committee, 1993).
3. Results and modelling 3.1. Gas concentrations Initial gas concentrations were measured approximately 30 min after dosing. The observed concentration was found to be within 5% of the targeted concentration, generally exceeding the target concentration due to the gas addition being calculated on an empty desiccator. No loss due to sorption was observed in the 10 and 158C fumigations; at the higher temperatures the head space concentration fell slightly to approximately 90% of the initial concentration over 7 days. For the purposes of modelling, the average concentration for each exposure was used. 3.2. Adults The survival of S. oryzae adults exposed to COS varied with concentration, time and temperature. Levels of carbon dioxide in the fumigation chambers rose over the duration of the exposure. In chambers where the level of kill was greater than 40%, the level of carbon dioxide did not exceeded 2%. Where mortality was low, fumigations were terminated when the concentration of carbon dioxide reached 4%. This limited the time of exposure for low doses of COS (less than 10 g m3). The survival for each concentration/time combination tested at 10, 15, 20, 25 and 308C is shown in Fig. 1. At each temperature, progressively lower concentrations took longer to achieve the same level of mortality, suggesting that an interaction between concentration and time is determining the level of mortality. The increasing survival at lower concentrations is suggestive of a threshold concentration, below which there is no effect on survival. The simplest model for the action of fumigant gases is that the product of the concentration (C) and the exposure time ðtÞ produces a constant level, i.e. Ct ¼ k of kill k (Haber, 1924), which has become known as Haber’s Rule. There has been considerable debate about the usefulness of Haber’s Rule and for what gases and under what conditions it applies. However, given small ranges for each of concentration, time and temperature, Haber’s rule could apply for most fumigant gases. While the simplicity of such a model is tempting, it could severely limit the range in which a fumigant is deemed useful. A model for mortality will identify the full range of Ct combinations at which this fumigant can be used and identify the limitations and conditions where fumigation may fail to kill all insects and/or select for tolerant insects. To test the adequacy of Haber’s rule as a model for these data, the term lnðCtÞ was fitted. This basic model was improved by the addition of subsequent terms. Adding the term lnðtÞ substantially reduced the residual deviance; however, this function was not linear and the term was better represented by lnðt þ 9:12Þ. The constant 9.12 was determined by maximum
388
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
Fig. 1. Survival of adult Sitophilus oryzae fumigated with carbonyl sulfide at a range of fixed concentrations and times at 10, 15, 20, 25 and 308C.
quasi-likelihood when this term and its interaction were included in the model. This feature of the model has no apparent mechanistic or biological explanation. The terms in order of fit for the adult mortality model and the analyses of deviance are given in Table 1. The analyses of deviance tables give the relative magnitude of the improved fit due to
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
389
adding each new term to the model. Tests of significance of each term in the models are based on t-statistics; these show the significance of each term allowing for all the others. The analysis of deviance table gives the significance of the fitted term allowing for the previous terms fitted but ignoring the subsequent ones. This leaves the general model lnfp=ð1 pÞg ¼ a þ blnðCtÞ þ clnðt þ 9:12Þ;
ð1Þ
where p in the probability of death, a, b, and c are constants, C is concentration in g m3 and t is time in h. The estimation of the value of Ct which achieves p 100% mortality is lnðCtÞ ¼ ½ln p=ð1 pÞ a clnðt þ 9:12Þ =b:
ð2Þ
When p ¼ 0:95, 0.99 and 0.999, lnðp=ð1 pÞÞ ¼ 2:944, 4.595 and 6.907, respectively. Temperature effects are statistically significant but might be considered small for practical purposes. If temperature effects are ignored and all data pooled, the fitted model for mortality is lnfp=ð1 pÞg ¼ 18:864 þ 6:296 lnðCtÞ 3:815 lnðt þ 9:12Þ:
ð3Þ
This model is illustrated in Fig. 2. A theoretical curve for 99% kill (which is within the domain of the data) and an extrapolated curve for 99.9% kill are plotted against the experimental data. Comparing this combined model with those for individual temperatures (Fig. 3), it is clear that the general model is not a good representation of mortality observed at extreme values such as 158C. However, under most conditions, a concentration in excess of 15 g m3 for 48 h or longer will kill at least 99% of adult S. oryzae. To increase the level of kill to 99.9%, given the same concentration of 15 g m3, the predicted duration of fumigation is more than 96 h (4 d). If the duration of fumigation were required to be less than 48 h, the dose required for 99.9% kill would need to be increased to more than 22 g m3. Caution should be exercised in interpreting from the 99.9% model as this is an extrapolation from the model and goes beyond the domain of the data. The tailing of the model at each axis suggests that concentrations required to kill adults in less than one day are likely to be very high and that concentrations of the fumigant below 10 g m3 are unlikely to achieve a 99.9% kill of adults. Table 1 Analysis of deviance for terms used to create the model for adult mortality following carbonyl sulfide fumigation. C is the concentration of carbonyl sulfide in g m3, t is time in hours and T is the temperature in 8C Terms in order of fit
d.f.
Deviance
Mean deviance
Deviance ratio Fa
lnðCtÞ lnðt þ 9:12Þ T20 lnðCtÞ ðT 20Þ lnðtþ9:12Þ ðT 20Þ Residual Total
1 1 1 1 1 176 181
5778 3280 65 269 228 793 10,413
5778 3280 65 269 228 4.51 58.53
1282 727 14 60 51
Adjusted R2 = 92.3%. a All t-values have P50:001 except for t=1.21 for which P=2.226.
390
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
Fig. 2. Theoretical curves for 99 and 99.9% kill, generated from the mathematical model for adult mortality for the temperature range 10–308C, plotted against observations of 100% (*), 95–99.9% (&), and 85–94.9% ( ) kill. For comparison, a plot of Ct ¼ 1284 g h m3 is included.
Fig. 3. Theoretical curves for 99% kill of adult Sitophilus oryzae fumigated with carbonyl sulfide at 10, 15, 20, 25 and 308C.
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
391
For comparison, a plot for best fit of the data to a Ct ¼ k model for 99.9% kill is included in Fig. 3 (Ct¼ 1284 g h m3 ). Comparison of the model to this plot shows that there is considerable variation between the generated model and a straight Ct model. If a Ct model were applied, fumigations undertaken for longer than 120 h periods would be at risk of underdosing and thereby increase the likelihood of survival and the ultimate development of COS-resistant insects. If temperature ðTÞ is included in the model, the major contributory components are the linear effect of temperature and the linear component of its interaction with the terms lnðCtÞ and lnðt þ 9:12Þ. The fitted model, with 208C as the reference temperature, then becomes lnfp=ð1 pÞg ¼ 20:644 þ 6:296lnðCtÞ 3:887lnðt þ 9:12Þ þ 0:5974 ð4Þ ðT 20Þ þ 0:0316ðT 20ÞlnðCtÞ 0:1938ðT 20Þlnðt þ 9:12Þ Theoretical curves for 99% kill at each of the tested temperatures are plotted as concentration against time in Fig. 3. These plots show how the relationship between concentration and time change with temperature. It appears that higher doses require a shorter exposure period to kill the same percentage at higher temperatures. According to this model, a fumigation of 25 g m3 will take about 12 h to kill 99% at 308C and progressively longer as temperatures decrease, taking more than 24 h to achieve the same level of kill at 108C. This can be interpreted from the raw data by a comparison of the level of kill achieved at one concentration and one time across the temperature range. For example, a 24-h fumigation of 15 g m3 killed 99% at 308C, 98% at 258C, 97% at 208C, 95% at 158C and only 92% at 108C. Perhaps of more importance is the difference between the minimal dose to achieve 99% kill over longer fumigations. Using 96 h as an example, it is clear that the concentration required to kill 99% is much higher at 308C than at 108C. Details of the parameter estimates for fitting the main terms for the adult models are given in Table 2. 3.3. Eggs As with the adults, egg survival varied with concentration, time and temperature. The level of carbon dioxide in the fumigation chambers never exceeded 0.5%. The survival for each Ct combination at 10, 15, 20, 25 and 308C is shown in Fig. 4. Table 2 Parameter estimates for the model for adult Sitophilus oryzae mortality following carbonyl sulfide fumigation. C is the concentration of carbonyl sulfide in g m3, t is time in hours and T is the temperature in 8C Term
Estimate
SE
T176a
Constant lnðCtÞ lnðt þ 9:12Þ T20 lnðCtÞ ðT 20Þ lnðt þ 9:12Þ ðT20Þ
20.644 6.296 3.887 0.5974 0.0316 0.1938
0.779 0.224 0.178 0.0956 0.0260 0.0268
26.49 28.12 21.83 6.25 1.21 7.24
a All t-values have P50:001 except for t ¼ 1:21 for which P ¼ 0:226.
392
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
In a similar vein to the modelling that was undertaken on the adult data, Ct was assumed to be the major term and to test whether C and t had equal impact the terms lnðCtÞ and lnðtÞ were used. Arbitrary smooth curves (splines) were fitted to see if there was any residual curvature; the splines
Fig. 4. Survival of Sitophilus oryzae eggs fumigated with carbonyl sulfide at a range of fixed concentrations and times at 10, 15, 20, 25 and 308C.
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
393
were almost straight, which indicated that log terms were suitable for use in modelling the data. Slight curvature in lnðtÞ was found which was reduced by adding a constant, for consistency the same constant of 9.12 (from the adult model) was used. Thus the term became lnðt þ 9:12Þ. Models for eggs that assume that each temperature level is different were analysed in Genstat using the Genstat procedure ‘probitanalysis’. To fit a model to all temperatures from 15 to 308C a novel Genstat code ‘fitnonlinear’ was used. The analyses of deviance for the main terms of egg mortality model are given in Table 3. Eq. (1) was again used as the basic logit model for mortality. Temperature, when fitted, was treated as a factor, because the linear term T 20 did not produce an adequate fit. Table 3 Analysis of deviance for terms used to create the model for egg mortality following carbonyl sulfide fumigation. C is the concentration of carbonyl sulfide in g m3, t is time in hours and T is temperature in 8C Terms in order of fit
d.f.
Deviance
Mean deviance
Deviance ratio F a
ln(Ct) ln(t þ 9:12) T lnðCtÞT lnðt þ 9:12ÞT Residual Total
1 1 3 3 3 88 99
4868 256 847 34 197 343 6544
4868 256 282 11 66 3.90 66.10
1248 66 72 2.90 16.92
a
All F-values have P50:001 except for F ¼ 2:90 for which P ¼ 0:039.
Table 4 Estimated parameters ( SE) for constants (a, b and c) in the model ‘‘logitðpÞ ¼ a þ blnðCtÞ þ clnðt þ 9:12Þ’’ which describes the mortality of Sitophilus oryzae eggs fumigated with carbonyl sulfide. Where p is the probability of death, C is the concentration of carbonyl sulfide in g m3, t is time in hours, n is estimated control emergence, and j is the variance scale parameter. The last row fits common values of a, b and c for all temperatures in the range 15–308C Temp.8C
n
a
b
c
j
10
562.1 ( 26.6) 368.2 ( 12.7) 286.8 ( 8.18) 126.8 ( 5.31) 351.9 ( 17.2) Various
29.214 ( 3.434) 43.000 ( 6.183) 36.041 ( 3.473) 37.469 ( 4.657) 29.373 ( 3.542) 23.665 ( 2.384)
2.703 ( 0.330) 3.538 ( 0.568) 4.454 ( 0.495) 5.189 ( 0.659) 4.778 ( 0.662) 2.941 ( 0.355)
2.121 ( 0.304) 3.889 ( 0.650) 1.499 ( 0.270) 1.160 ( 0.265) 0.180 ( 0.367) 1.037 ( 0.263)
10.62
15 20 25 30 15–30
5.413 2.321 1.773 6.535 14.62
394
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
Fig. 5. Theoretical curves for 99% kill of Sitophilus oryzae eggs generated from the mathematical models for egg mortality. A separate curve is given for each temperature 10, 15, 20, 25, 308C, and the combined temperature range 15–308C.
Table 5 Time to kill 99.9% of Sitophilus oryzae eggs (calculated from the individual temperature models), when fumigated at a rate of 25 g m3 carbonyl sulfide, at 10, 15, 20, 25 and 308C Temperature (8C)
Concentration (g m3)
Time (h)
Approximate time (d)
10 15 20 25 30
25 25 25 25 25
290 174 120 77 95
11 7.5 5.0 3.2 4.0
Details of the parameter estimates for fitting the main terms for the egg model for each temperature are given in Table 4. Separate models for 99% kill fitted for each temperature are compared with that fitted for all of the data from 15 to 308C in Fig. 5. From these theoretical curves it appears that COS is most effective in the range of 20–308C. Fumigation at 15 or 108C takes longer to achieve the same level of kill given the same concentration. The longer duration could be due to a reduction in metabolism or slowing of development at lower temperatures. Not surprisingly the model for the combined temperatures (15–308C) most closely mimics the 158C curve. The requirement for longer fumigations at 158C pushes the time required to achieve 99% kill out to more than 96 h (4 d) at 25 g m3, as opposed to about 48 h for 25 g m3 at 258C. As was the case with the adult model, the combined temperature model is not a good representation of mortality observed at
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
395
Fig. 6. Theoretical curves for 99 and 99.9% kill, generated from the mathematical model for egg mortality for the temperature range 15–308C, plotted against observations of 100% (*), 95–99.9% (&), and 85–94.9% ( ) kill. For comparison, a plot of Ct ¼ 4512 g h m3 is included.
extreme values such as low concentrations and low temperature (158C). For instance, if a dose rate of 15 g m3 was chosen for the fumigation of S. oryzae eggs from the general model, and applied at 158C, there could be strong selection for tolerant insects and a likelihood of emergent resistance to COS. It is therefore preferable to apply COS according to the individual temperature models rather than relying on the combined temperature model. Extrapolating from the individual temperature models, Table 5 gives the theoretical times taken to kill 99.9% of S. oryzae eggs with a concentration of 25 g m3 at 10, 15, 20, 25 and 308C. An extrapolated plot for 99.9% kill from the combined temperature model is compared with the experimental data in Fig. 6. Although 99.9% kill is beyond the domain of the data collected in this experiment it gives an idea of how much longer a fumigation will need to be to achieve the increased level of kill. For comparison, a plot for best fit of the data to a Ct ¼ k model for 99.9% kill is included in Fig. 6 (Ct¼ 4512 g h m3 ). In this instance a basic Ct model would underestimate the dose required to achieve 99.9% kill up until 264 h after which the dose would be overestimated. 4. Discussion From the observed mortality of S. oryzae eggs and adults, fumigated with COS, mathematical models have been developed which can be used to predict the concentrations and times
396
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
required for various levels of kill over a range of temperatures. Now the debate can focus on what level of kill is acceptable and safe in terms of avoiding the possible development of resistance. Fumigations to kill only adults (and larvae) are not recommended for the long-term control of insect pests, but may be applied under certain conditions, or as a sequence of fumigations a couple of weeks apart, allowing for the development of the more tolerant life stages to a more susceptible stage. The models developed in this paper for the control of S. oryzae adults indicate that within the temperature range of 15–308C, a 24 h exposure at 30 g m3 will kill 99% and that concentrations below 10 g m3 are unlikely to result in complete control of S. oryzae even with extended exposures such as 10 to 14 days. Fumigations to control all life stages of pests are routinely undertaken and setting an appropriate concentration and time regime for the fumigation is vital. S. oryzae eggs were more difficult to kill with carbonyl sulfide fumigation than the adults. At 308C, a 25 g m3 fumigation is predicted to kill 99.9% of adults in less than 1 d, but will theoretically take 4 d to kill the same percentage of eggs. The models developed for the control of S. oryzae eggs at each temperature emphasise the importance of understanding how environmental factors such as temperature can affect the outcome of a given fumigation. Something as simple as 58C temperature change can alter a proposed 5 d fumigation at 208C (25 g m3 to achieve 99.9% kill) to a 7.5 d fumigation at 158C (to achieve the same level of kill from the same concentration). That is another 2.5 d to maintain the concentration and withhold the goods. Temperature is only one of the considerations that can affect the outcome of fumigations. Many other factors have not been considered, including sorption by the commodity and other target species. Although it was assumed that S. oryzae eggs are the most tolerant life stage of one of the more tolerant species of stored product insects (Desmarchelier, 1994; Tan et al., 1999; Plarre and Reichmuth, 1997; Zettler et al., 1997; Weller, unpublished data), it is possible that a species not yet tested against COS will be more tolerant. With this in mind, it should not be assumed that the models generated for the use of COS in this paper will be applicable to all insects. In summary, COS has potential as an alternative to both methyl bromide and phosphine for fumigation of stored products. Although the Ct product required for a successful fumigation with COS is much higher than is typical for methyl bromide (approximately 2000 g h m3 for COS compared with 200 g h m3 for methyl bromide), the higher Ct product is achievable in practice because of lower sorption of COS than methyl bromide onto structures and commodities (Desmarchelier, 1994; Ren, 1996). Unfortunately, fumigations undertaken with COS will be significantly longer that those with methyl bromide but will be shorter to achieve the same level of kill than with phosphine, carbon dioxide or low oxygen (nitrogen).
Acknowledgements The research reported in this paper was supported by funds from the Partners to the Stored Grain Research Laboratory Agreement and CSIRO Australia. The authors are also grateful to Jim Desmarchelier, Peter Annis, Jonathan Banks, Albert Trajstman and Yong Lin Ren for their critical review and comments of this manuscript.
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
397
References Banks, H.J., 1994. Fumigation } an endangered technology? In: Highley, E., Wright, E.J., Banks, H.J., Champ, B.R. (Eds.), Stored Product Protection, Proceedings of the 6th International Working Conference on Stored-Product Protection, Canberra, Australia, 17–23 April 1994. CAB International, Wallingford, Oxon, UK, pp. 2–6. Banks, H.J., Desmarchelier, J.M., Ren, Y.L., 1993. Carbonyl sulfide fumigant and method of fumigation. International Application Published under the Patent Cooperation Treaty (PCT), WO93/13659. Bengston, M., Acada, M., Daglish, G.D., Collins P.J., 1999. Phosphine resistance in the Asia/Pacific region. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 631–634. Bui Cong Hien, 1999. Some initial results on phosphine resistance of major product insect pests in Vietnam. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 648–652. Desmarchelier, J.M., 1994. Carbonyl sulfide as a fumigant for control of insects and mites. In: Highley, E., Wright, E.J., Banks, H.J., Champ, B.R. (Eds), Stored Product Protection, Proceedings of the 6th International Working Conference on Stored-Product Protection, Canberra, Australia, 17–23 April 1994. CAB International, Wallingford, Oxon, UK, pp.78–82. Fields, P.G., 1992. The control of stored-product insects and mites with extreme temperatures. Journal of Stored Products Research 28, 89–118. Finney, D.J., 1980. Probit Analysis, 3rd ed. Cambridge University Press, Cambridge. Genstat 5 Committee, 1993. Genstat 5 Release 3 Reference Manual. Clarendon Press, Oxford. Haber, F., 1924. Fu¨nf Vortra¨ge aus den Jahren 1920–1923. Springer, Berlin, p. 98S. Ignatowicz, S., 1999. Cases of phosphine resistance for grain weevil, Sitophilus granarius found in Poland. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 625–630. McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models, 2nd ed. Chapman & Hall, London. Plarre, R., Reichmuth, C., 1997. Effects of carbonyl sulfide on Sitophilus granarius (L.) (Coleoptera: Curculionidae), Fusarium culmorum and Fusarium avenaceum (SACC.) (Deuteromycotina: Hyphomycetes), and corrosion on copper. In: Donahaye, E.J., Navarro, S., Varnava, A. (Eds.), Proceedings of the International Conference on Controlled Atmosphere and Fumigation in Stored products, Nicosia, Cyprus, 21–26 April 1996. Printco Ltd., Nicosia, Cyprus, pp. 59–71. Rajendran, S., 1999. Phosphine resistance in stored grain insect pests in India. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 635–641. Ren, Y.L., 1996. Carbonyl sulfide as a fumigant for grain and timber. Ph.D. Thesis, University of Canberra, Australia. Tan Xianchang, Hou Xingwei, Cheng Lizheng, Wu Jianchun, 1999. Research on carbonyl sulfide as a fumigant for control of stored grain insects. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 567–571. U.N.E.P., 1996. Eighth Meeting of the Parties to the Montreal Protocol on Substances that Deplete the Ozone Layer. Vienna, November 1996. Zeng, L., 1999. Development and counter measures of phosphine resistance in stored grain insects in Guangdong of China. In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product
398
G.L. Weller, R. Morton / Journal of Stored Products Research 37 (2001) 383–398
Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 642–647. Zettler, J.L., Leesch, J.G., Gill, R.F., Mackey, B.E., 1997. Toxicity of carbonyl sulfide to stored product insects. Journal of Economic Entomology 90, 832–836. Zettler, J.L., Leesch, J.G., Gill, R.F., Tebbets, J.G., 1999. Chemical alternatives for methyl bromide and phosphine treatments for dried fruits and nuts In: Jin Zuxun, Liang Quan, Liang Yongsheng, Tan Xianchang, Guan Lianghua (Eds.), Stored Product Protection, Proceedings of the 7th International Working Conference on Stored-Product Protection, Beijing, China, 14–19 October 1998. Sichuan Publishing House of Science and Technology, Chengdu, Sichuan Province, P.R. China, pp. 554–561.