Cotton Boll Weevil: An Evaluation of USDA Programs : a Report (1981)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5. APPRAISAL OF THE OPM AND BWE TRIALS AND PLANS FOR THEIR BELTWIDE APPLICATION INTRODUCTION The OPM and BWE trials were large-scale demonstrations, and several constraints made it impossible to plan them scientifically. In the case of the BWE trial, one constraint was the need for a large buffer zone that sufficiently isolated the trial area from other weevil-infested areas. Otherwise, suppression of the population in the trial area could be obscured by in-migration from a non-trial area. It was this constraint that led to the selection of areas in North Carolina and Virginia for the BWE trial. As a result, the BWE trial was conducted in areas where the boll weevil population was at a very low density. The need for a large-scale buffer zone also resulted in cost constraints that limited eradication to a single trial. This unreplicated trial in an area untypical of the Cotton Belt as a whole prevented any statistical analysis of the BWE trial's beltwide feasibility. The OPM trial was also unreplicated and so limited in size that in-migration of boll weevils from adjacent areas was clearly detected. There were two criteria used by the USDA for success in the OPM and BWE trials (USDA l98lb): (l) Demonstrate proof of the biological success, environmental acceptability, and economic feasibility of each method; (2) Supply a data base for making judgments on the feasibility of beltwide implementation. A reliable test of the success of the BWE trials depended on the establishment of statistically acceptable confidence intervals for two factors: (a) pheromone trap efficiency, and (b) log rate of population growth for the boll weevil. Since an eradication program, by definition, deals with populations driven to the verge of extinc- tion rather than with populations in normal ranges, the values obtained for both of these factors may have been quite different than they would have been for normal populations. If both of these fac- tors are known, at any rate, the trap density needed to assure detec- tion of any surviving weevils can be calculated. 58

59 There were three possible outcomes for the BWE trial: (l) a measurable number of boll weevils after the trial period ended would mean that eradication had failed; (2) no detectable weevils in the trial area would mean that eradication was successful at a given detection level; (3) detection of low numbers of boll weevils could mean either an in-migration of weevils to the test area or low levels of uneradi- cated weevils in the test area with a low rate of population increase. The USDA's biological evaluation team report (SEA l98l) did not give this last possibility the same degree of consideration as in- migration. As for the OPM trial, there were many uncontrolled variables that make it difficult to analyze the trial's success. There were inherent differences in the weather and the fertility of the fields in Panola and Pontotoc counties. There were also differences in the numbers of secondary insect pests (e.g., more tarnished plant bugs in Pontotoc in l979). Late-season beneficial insects tended to be higher in Pontotoc, and there were differences in the amounts and kinds of insecticides used in the two counties. In-season spraying against Heliothis, for example, was higher in Panola. As originally envisioned, the BWE trial would have used sterile male boll weevils as the keystone of the eradication process. But the development of methods to produce large numbers of highly competitive, fully sterile boll weevils has proved to be a highly intractable problem. There is no evidence that sterile boll weevils contributed significantly to population suppression in the BWE trial. Furthermore, some USDA scientists have even suggested that insecti- cides (a density-independent mortality factor) can serve as a substi- tute for sterile males (a density-dependent mortality factor). Two years of surveillance without a single find raises the probability of boll weevil eradication in North Carolina's Chowan County, which was part of the BWE trial area (SEA l98l). Two argu- ments can be made against that probability: (l) non-detection does not necessarily mean extinction, and (2) some earlier claims about the eradication of certain insects have been erroneous, since relict populations were found years later (Michigan Department of Agricul- ture l965, NRC l969, Wallner l974). Until a considerable period of time has elapsed, therefore, eradication cannot be demonstrated with reasonable certainty. The situation in Chowan County is best described as a two-year non-detectable weevil population at l phero- mone trap per acre monitoring density. A description of the situa- tion in the other counties of the eradication zone would be one weevil per l0 thousand acres of cotton detectable at l pheromone trap per acre monitoring density. Since North Carolina and Virginia are at the northern limit of the boll weevil's range and weevil populations in the trial zone had been reduced to extremely low levels by two unusually cold winters immediately before the trial, it would be expected that the APHIS program would reduce boll weevil populations more rapidly in North Carolina and Virginia than could be anticipated elsewhere in a belt- wide program. A more vigorous program of population suppression

60 would be needed in other areas of the Cotton Belt where much larger populations of the boll weevil are entrenched in a more favorable climate. A more reasonable assessment of the BWE trial is that an intensive program of three years' duration followed by two years of intensive monitoring would be the minimum needed to reduce the popu- lation to a non-detectable level throughout the Cotton Belt. Beltwide plans must take into account the subtle interactions of multiple factors, and the chances for error in extrapolating from the trial area to beltwide are obvious. In order to decide which program to adopt, the estimate of benefits should significantly exceed any margin of error in making these estimates. Since the OPM and BWE trials were unreplicated demonstrations, lacked suitable controls, and involved uncontrollable variables, the scientific knowledge necessary to determine their degree of success or failure is limited. Any assessment of their value or extrapola- tion from them is a matter of individual judgment. BIOLOGICAL CONSIDERATIONS The Committee has evaluated the USDA biological evaluation team's report and commends its members for a job well done, given the constraints imposed. The shortcomings of the report result from the design of the trials, which did not provide the necessary data from which to draw conclusions applicable to the entire Cotton Belt. These shortcomings give rise to major areas of concern, which are discussed below. Homogeneity of Eradication Trial Versus Heterogeneity of Cotton Belt Cotton is grown over a wide range of soil and climatic condi- tions, and is attacked by a wide variety of insects. Thus, no one group of pest control practices can be recommended for all cotton- growing areas. A major purpose of the eradication trial in North Carolina was the need for information to evaluate the biological implications of a boll weevil eradication attempt. We found that the BWE trial had provided insufficient data and information to resolve several signif- icant issues. The trial did not provide information useful for boll weevil eradication throughout the Cotton Belt, nor was it designed to do so. We discuss here the subjects on which little or no informa- tion exists and the research that should be undertaken on these subjects. The eradication trial was conducted in a relatively homogeneous environment, in contrast to the heterogeneous environment of the entire Cotton Belt. This heterogeneity extends to the boll weevil itself. Hence, one important question is, How would the boll weevil respond to a uniform eradication effort, as proposed in the BWE plan? This question leads to more specific concerns:

6l • The boll weevil population in the BWE trial area was at a record low. Thus, it might normally take a year longer to reduce average boll weevil populations to the level achieved in the trial. At the very least, therefore, the cost esti- mates for a beltwide BWE program should be increased by the costs of one additional year. • The effect of climatic conditions on boll weevil populations varies across the belt. Would these differences affect the eradication program? • Do all weevil populations respond in the same way to phero- mone traps? Is the pheromone trap equally efficient in all of the microclimates that would be included in the beltwide eradication program? Experience with other pheromone traps, such as those developed for the European corn borer, has shown that insect populations can vary widely in their res- ponse to a given synthetic pheromone. • Would boll weevil populations respond equally to the same amount of insecticide, or do they possess varying levels of resistance due to differences in previous exposure to insect- icides? Environmental differences across the Cotton Belt may also substantially influence the success of an eradication effort. Al- though insect complexes vary substantially across the belt, regional differences in rainfall and other climatic factors have been even more important in determining the best methods of managing cotton insects. Would not these same differences influence the effective- ness of a uniformly applied eradication effort? Differences in temperature and humidity may influence the effectiveness of traps and insecticides as well as the behavior and reproductive potential of the weevil. It is already known that differences in the amount of suitable overwintering habitat, soil conditions, and cropping pat- terns influence insect management practices. In Texas, for example, diapause control has proved effective in suppressing weevil populations in west Texas, but diapause programs on a lesser scale in south Texas have met with limited success (Rummel and Frisbie l978). The environmental characteristics of south Texas—mild winters and abundant overwintering habitats—are more favorable for overwintering weevils than the harsher conditions of west Texas. Since the south Texas environment is not unlike that in other southern cotton-producing areas, the effectiveness of dia- pause insecticide applications as a component of the eradication program seems debatable. The BWE trial did not offer the opportunity to evaluate this potential problem. Different cotton varieties are planted across the Cotton Belt, but no measure of how these varietal differences might influence the effectiveness of an eradication effort is in hand. Variations in

62 planting and harvesting dates, stalk destruction, and other cultiva- tion practices also might influence the outcome of the eradication program. Another problem is that there are host plants other than cotton in the United States that can sustain boll weevil populations (SEA l98l) . One of these is Ciefuegosia drummondi, a wild host found in the Coastal Bend region of Texas, and another is Hibiscus. In addi- tion, several malvaceous plants listed as potential hosts occur in southern Florida, although these are presumed to be too far from present cotton-growing areas to present much of a problem. There is also the possibility that cotton plants in urban areas grown as curiosities or ornamentals could harbor low boll weevil populations, as could cotton plants growing along roadways or other sites where seeds were lost from farm equipment. Any alternative host species must be carefully evaluated and, if found, the sources or the weevils eliminated. Another thing that should be mentioned is that a large array of pest control practices are currently used throughout the belt. If an eradication program replaced these current practices, the response of insect species other than the boll weevil could vary. Some might change from non-pest or occasional pest to key-pest status. Further- more, these insect species might vary from location to location in their ability to develop resistance to insecticides. In summary, considerable heterogeneity exists across the Cotton Belt in terms of the cotton insect complex, environmental character- istics, cotton production practices, current insect management prac- tices, and perhaps in boll weevil populations as well. The BWE trial provided no information on how effective its techniques would be in dealing with this heterogeneity. We have emphasized heterogeneity with respect to an eradication program because differences at near-zero insect population levels are more important than differences at normal or managed population levels such as would occur under OPM. The eradication plan does not take into account the heterogeneity described above, unlike programs now in use. Furthermore, each state in the Cotton Belt has now developed a modification of the basic OPM strategy to fit local conditions (Economics and Statistics Service l98lc). Although none of the states has tested its program on as large a scale as the OPM trial, they do have a plan which attempts to deal with heterogeneity. Pheromone Traps As stated earlier in Chapter 2, grandlure is especially useful in boll weevil detection surveys and in predicting the need for insecticide applications. Further research may demonstrate that boll weevils emerging from diapause in the spring may be effectively "trapped-out" with pheromone traps.

63 Use of Traps in the OPM Trial Pheromone traps were used in the OPM trial to provide estimates of boll weevil populations. In the spring and in the fall one trap per l5 to 20 acres was installed infield and peripherally throughout all the cotton acreage in Panola County and in the fall in nearby Pontotoc County. Traps were inspected and serviced weekly. The data on the numbers of weevils captured were extrapolated to give esti- mates of the number of boll weevils per acre. These extrapolations were based upon earlier studies on the effectiveness of the traps. In the OPM trial the traps gave warning of overwintering weevils leaving diapause in numbers sufficient to require insecticide treat- ment in early season. Estimates of boll weevil population increases during the season were also obtained through the use of traps. Use of Traps in the BWE Trial Pheromone traps provided useful information during several phases of the BWE trial, thus making it possible to determine boll weevil population levels to a degree of accuracy never before possi- ble. Furthermore, their ability to capture weevils at extremely low population densities simplified the program, since data were available to assist in making decisions. Post-eradication monitoring would also be simplified by traps, since the need to scout for weevil survivors or immigrants would be reduced and perhaps eliminated except in special situations. Pheromone traps would be used at several stages in the beltwide eradication plan. Pheromone Traps: Some Unanswered Questions The NRC Committee recognizes the sophisticated research that has been done on the boll weevil pheromone, including the accomplishment of developing methods for using the pheromone in a practical way. In several respects the pheromone traps are the foundation on which a beltwide BWE program would be based, since the traps would be essen- tial to surveying, predicting, and monitoring the boll weevil popula- tion. Some of the following comments are probably minor, but the traps are so important to the program that their failure could result in significant setbacks. Potential Variations in Weevil Response to Traps. It has been demonstrated with the pine engraver beetle (Ips pini) that a signifi- cant differential response to pheromones exists over its range of distribution (Lanier et al. l980). Such a differential response in boll weevils may not exist. Grandlure has been studied widely for its usefulness throughout the infested areas of the Cotton Belt, and there is little to suggest that boll weevil populations are geneti- cally isolated from one another. Nevertheless, even slight varia- tions in the response of boll weevil populations to the pheromone could have a profound impact upon trap density, distribution, and

64 overall program design. This may warrant careful evaluation and additional research. Insect behaviorists have speculated about variations in the response to pheromone traps of individuals within a single population. If significant variation exists, pheromone traps may exert a selec- tion effect on the boll weevil population for those weevils that are least attracted to the traps. Variations could be a reaction to the fluorescent yellow color of the trap as well as to the pheromone itself. It seems unlikely that such variations would occur rapidly, if at all, but the possibility of unanticipated responses should be carefully evaluated. Trap Efficiency and Density. The pheromone traps are remarkably efficient, and it has been clearly established that efficiency is inversely related to density. Thus, efficiency is better when popu- lations are low, as would be the case in an eradication campaign. It is predicted that one trap per acre would be almost l00 percent cer- tain to discover the F2 generation of any infestation left behind in an eradication program or initiated by an immigrant. The NRC Committee is concerned, however, about the proposed trap density, particularly during the monitoring stage, although we realize that research in the field to determine the exact trap density needed is probably impossible. What concerns the Committee is that trap den- sity may be too low in any case. Insect populations frequently do not behave as expected. Because of weather conditions, for example, they may not increase normally. Hot, dry summers and excessively cold winters are inimical to normal population increases. In extreme situations it might take many generations, or even years, for a suppressed population or a reinfestation to reach detectable levels at the planned trap density. Effects of Weather on Pheromone Trap Efficiency. In the case of some insects, weather factors are considered to affect pheromone trap efficiency (Elkinton and Garde l980). Variables such as heat or cold, dryness or humidity, wind or no wind, and perhaps other factors may upset expected results, although these may be only temporary interruptions. The effect of weather on boll weevil trap efficiency has not been studied, insofar as the Committee is aware. Detectable Population Levels. Despite intensive research on boll weevil pheromones and trapping techniques, there is no clear understanding of the minimum population levels that would be detected. Low-level populations, which may exist after the eradication effort has moved through a zone, may not increase equally in all locations. Population increases may be very slow and may not follow the pattern observed during the original weevil invasion. Large amounts of insecticides are used on cotton for pests other than the boll weevil. These insecticides may change the rate of weevil increase. Rapid increase of ?2 and F$ generations is an essential assumption for early detection and required for a successful eradication program. Low trap density in the monitoring phase might result in large increases and wide dispersal of weevils prior to detection. If the latter occurred, the capture of few or no weevils for several years after eradication might be followed by a rapid increase of the boll weevil population over a widespread region.

65 Since damage to U.S. cotton by the boll weevil was discovered in Texas in the early l890s, it is possible that the insect had reached southern Texas counties from Mexico as early as the l870s and adapted itself to various environmental situations. It might therefore be l5 to 20 years before boll weevils adapt sufficiently to a new environ- ment to be able to reproduce in large numbers. Other insects have been observed to require a similar span of time between first detec- tion and economic damage levels. It has been estimated, for example, that it took l7 years after its introduction for the cereal leaf beetle to reach damaging population density (Haynes and Gage l98l) . Migration and Dispersal of the Boll Weevil A key element of a successful eradication program would be keeping zones freed of boll weevils isolated from zones still infest- ed. Our ability to do so may be complicated by the fact that there are two distinct ways in which the boll weevil may have invaded North America. One would have been the gradual buildup of an invading population in a new area, followed by further spread. This concept is used to justify eradication on grounds that it can be successfully maintained. An equal probability, however, is that the boll weevil population dispersed uniformly from its original location, with individuals settling into North American cotton as a function of the square of the distance travelled. Boll weevils may therefore be continually dispersing into North America irrespective of the native North American population. If this were the case, eradication would not be possible unless the original source of the weevils was also eliminated. We have no data to support or refute either argument. The information needed to judge between these two possibilities would have to come from population densities below detectable thresholds. However, a mathematical model could be constructed for both possibil- ities and perhaps tested by standard validation experiments. Cross (l98l) states that weevils make short flights that are not in response to pheromones. These dispersal flights are random and appear to be related only to wind and sun location. Boll weevil habits change in August, particularly in west Texas, as females search for better oviposition sites or diapause locations. These migrants have been known to travel as far as 80 kilometers. This long-range dispersal occurs when weevils on short flights are borne aloft by thermal wind currents. Such behavior is quite common among other Coleoptera. Raun (Pest Management Consultants, Inc., Lincoln, NE, personal communication, l98l) reports a large flight of corn rootworm beetles at 4000 feet above ground level over western Nebraska at about 5:00 p.m. in late July of l976. Flight-mill studies have shown the boll weevil to fly an average distance of 3.56 kilometers. The longest flight in these studies was l7.66 kilometers. If one speculates that weevils carried by prevail- ing southwesterly breezes also actively transport themselves, an 80-kilometer flight may not be uncommon. Davich et al. (l970) cap- tured weevils 40 to 72 kilometers from cultivated cotton.

66 Boll weevil migration and dispersal can occur by several means and for various reasons: • Weevil movement is stimulated by pheromone attraction, by the need for hosts upon which to feed and by the need to find egg-laying and overwintering sites. • Weevils also make random flights of short duration for no identifiable reason. • Weevils may be borne aloft and carried for a substantial distance by the wind. • Catastrophic climatic occurrences (tornadoes, hurricanes) can lift weevils or plant parts with weevils and carry them long distances. • There is the further possibility of weevil dispersal through automotive, rail, or aircraft transportation, either in bolls as larvae or as adults. The final report of the BWE trial (APHIS l98lb) indicates the following forms of weevil "reinfestation": (l) Two adults were trapped in l979. One was found near a motel, the other near an APHIS field office. Both were found in the fall and are believed to have been transported by vehicles from outside the evaluation zone. (2) A partly disintegrated adult was found in a trap in May of l980. This is believed to have been a weevil caught the previous year but not removed from the trap. (3) Two weevils were trapped in the evaluation zone on August l8, l980, and another was trapped September l5. All 3 were in traps at a distance from cotton fields. (4) On September ll, l980, one adult was trapped in a cotton field. Between September l5 and 24, l980, 9 boll weevils were detected by visual examination of cotton bolls, or by trapping, at the same site. Three of these 9 were inside unopened bolls, with one of them in the pupa stage. The BWE trial final report also indicates trap captures with the "migration zone" grid of traps. These captures extended as far as l44 kilometers from infested areas. Although boll weevil suppression in the eradication zone was outstanding, the capture of occasional weevils in both l979 and l980 and the discovery of at least one breeding site in l980 led the NRC Committee to speculate on these captures. The accuracy of this speculation is largely dependent on how efficient grandlure traps are, an efficiency not known. If we accept the view that the probability of detecting l weevil per acre with l trap per acre is 60 percent (Leggett et al. l98l), extrapolation would indicate considerable opportunity for incipient weevils to escape detection by the system used in the BWE trial. It

67 is also possible that the trapped and discovered weevils were natural migrants who were either carried in by winds, flew in on their own power, or were transported by vehicular traffic. Whichever explanation is accepted, only time will reveal whether the boll weevil has been eradicated from the evaluation zone. These weevil finds, however, raise questions left unanswered by the BWE trial that must be considered in any attempt to extrapolate the results of the trial on a beltwide basis. If these weevil finds indicate migration reinfestation or an incipient infestation that was not eradicated, a considerable problem would be posed. But if these individual weevils are just individual weevils and not indicators of a larger population, the problem is not nearly as great. Nonethe- less, individuals and small populations that escaped a beltwide eradication program would have to be discovered immediately. If they went undetected, they could remain undetected for several generations by the proposed trap net. Various estimates of the reproductive potential of the boll weevil have been given. Most of them are qualified and, depending on environmental conditions, range from below zero to as high as 80x per generation. The most common increase reported per generation seems to be 2x to 8x. With this level of reproduction, and a trapping efficiency of only 60 percent (when traps are used at a level of l per acre), the maintenance trap net could easily fail to disclose breeding populations. This would be particularly true during a period when environmental conditions kept reproductive efficiency low for a period of several years. Quarantine methods are proposed to prevent migration and dis- persal. These might be effective for weevils on vehicles but would fail in the case of wind-borne or flying migrants. The movement of the eradication frontier, from east to west, is in the direction most likely to allow leap-frogging of the frontier. The direction of prevailing winds would be from infested areas into eradicated areas. This would also be true at the Mexican border, with migrants riding air currents across the buffer zone. The NRC Committee believes that the USDA's biological evaluation team should have considered various scenarios before concluding that the BWE trial had accomplished its objectives. If the team consid- ered other scenarios for explaining nondetection of weevils in the eradication zone and discarded them as unlikely, it should have discussed its rationale for doing so. Role of Sterile Males in Eradication Conclusions about the efficacy of sterile male boll weevils in suppressing low-level populations to the level of non-detectability rest upon (l) an experiment performed in Louisiana in l962 in which an induced low population of weevils was reduced to non-detectability by the end of the season (Davich et al. l965) and (2) theoretical considerations of the over saturation of low-level populations with sterile weevils based on information from experiments which failed to

68 achieve their objective but provided some information. It is on this basis that estimates have been made as to what the results of care- fully controlled field operations might be. The lack of conclusive data is disturbing, and two basic ques- tions remain to be answered. Can eradication be consistently achieved in different areas by oversaturation with sterile weevils? If so, what level of oversaturation is necessary to achieve it? The answers to both are theoretical and controversial. The role of sterile males in the proposed BWE program is unclear. The costs for operators and for the dispersal of sterile weevils are included in the estimates, but capital expenditures for the necessary expansion of facilities to produce the sterile males are not. Comments by USDA personnel as to the necessity for a ster- ile male component in the proposed BWE program were conflicting. Effect of Boll Weevil Eradication on Biological Control of Heliothis Insecticide Use A reduction in future insecticide use has been identified as a major justification for a beltwide eradication program (USDA l98lb). If the amount of insecticide used in current insect control is taken to be l00 percent, estimates of insecticide use in an OPM-NI program and an OPM-NI-BWE program are 77 percent and 54 percent, respectively. In the NRC Committee's view these estimates, derived from Delphi process estimates, are questionable. Judging by the past history of efforts to eradicate such insects as the gypsy moth and fire ant, insecticide use increases during the eradication program, and it is impossible to establish the time at which eradication will be accomplished. It has been repeatedly stated that frequent insecticide applica- tions for boll weevil control have produced "biological deserts" virtually devoid of natural parasites and predators. As a result of being freed of normal environmental hazards, the bollworm and the tobacco budworm have become rampant pests requiring ever more fre- quent insecticidal treatment. Because of the rapid onset of Helio- this resistance to insecticides (see Table 2.l) the situation has worsened and led in some areas to the total elimination of cotton cultivation (Reynolds et al. l975). It is by no means certain that the removal of the boll weevil would restore the ecological balance. Thirty years of genetic selec- tion and recombination in Heliothis due to intensive insecticide use have produced resistant biotypes of these pests that are considerably different from their progenitors. Beneficial insect populations in cotton fields have been reduced because of heavy insecticide use. Moreover, cotton varieties, cultivation practices, and insecticidal treatments on adjacent crops are entirely different from those of 30 years ago. The use of massive insecticidal applications for cotton pest control is solidly ingrained in cotton growers. It is therefore

69 quite possible that essentially the same amounts of insecticides will continue to be used, regardless of the presence or absence of the boll weevil. The estimates of insecticide use in OPM-NI and OPM-NI-BWE pro- grams also underestimate the potentialities of CIC as it evolves into integrated pest management in which ecologically based population regulation strategies are guided by skilled entomologists in the private sector operating under principles now under development by inter-university consortia. Such endeavors have already achieved major reductions in insecticide use in such areas as the Coastal Bend in Texas (Phillips et al. l980). The most conservative estimates of the effectiveness of IPM programs suggest reductions of 50 to 75 percent in current insecticide use. Thus, over time, current insect- icide use may decrease materially from the l00 percent level upon which the OPM-NI-BWE program is based and justified (USDA l98lb). Of major concern in this regard is the level of federal and state sup- port for integrated pest management that can be expected if massive USDA expenditures are made for OPM-NI-BWE or 0PM programs. Host Plant Resistance (HPR) Cotton varieties have a number of morphological and biochemical characteristics that confer measurable resistance to insects, and the beneficial insect complex of parasites and predators tends to moder- ate pest population levels. Therefore, beneficial arthropods and host plant resistance are complementary elements in crop protection. Only a few of the known host-plant resistance factors (Table 5.l) have been incorporated into commercial varieties of cotton. Most of these factors also have a negative effect that reduces yield when the genes controlling these factors are transferred into a commercial variety. Pubescent cotton, which has a positive breeding effect, and nectariless cotton, which is neutral, are exceptions to this general rule. Good varieties carrying these traits are already in use. The effect of the remaining HPR factors is cumulative, in that yield decreases with each additional HPR factor transferred. This limits the number of HPR factors that can be added to a commercial variety. Another problem develops when an added HPR factor causes susceptibility to a different insect species. Heavy pubescence, for example, causes a cotton plant to be resistant to Lygus, fleahopper, and leafhopper attack but susceptible to Heliothis. If the boll weevil were eradicated, plant breeding efforts could be targeted against either Heliothis, pink bollworm, or Lygus. Nectarilessness is available to offer some protection against all three. Other HPR factors could be added on a regional basis, depend- ing on which local pest causes the most severe problem.

70 TABLE 5.1 Host plant resistance (HPR1 factors and their relative effect on cotton yield and susceptibility to key cotton insect pests. Effect of HPR Factor on Boll Plant weevil Heliothis bugs Pest Species Pink bo ll worm HPR Factor Effect on yield Pubescent Increase - + - 0 Glabrous Decrease 0 - + - Nectariless Neutral 0 -? - High gossypol Decrease 0 0 High tannin - 0-0 0 Red plant (R^) Decrease 0 0 0 Red stem (R2) Decrease 0 0 0 Frego bract Decrease -? + 0 Okra leaf Decrease 0 0 0 Male sterile Decrease - 0 0 0 Oviposition suppression factor Unknown 0 0 0 AET 5 Unknown 0-0 antibiosis - « Less damage than normal + = More damage than normal 0 = No significant effect documented ? = Controversial results

7l Beneficial Arthropods Cotton fields contain a surprisingly varied and complex insect pest and entomophagous fauna. Many entomologists are convinced that it would not be possible to produce cotton economically without parasitic and predaceous insects (Reynolds et al. l975). Changes in cotton cultivation techniques (e.g., fertilizer and irrigation) and the use of different cotton varieties should be carefully evaluated for more than total yield and quality of fiber and seed before being adopted. Such changes can have profound effects upon beneficial insects as well as insect pest populations. Applications of chemical insecticides, particularly when they are repeated, often reduce the numbers of beneficial insects, largely negating their impact. As a consequence, populations of "primary target" species may resurge rapidly following insecticide applica- tions, and secondary pests also may increase. Both the BWE and OPM trials demonstrated better management of beneficial arthropods. Careful management and timing of insecticide applications to control boll weevils reduced insecticide loads. As a result, Heliothis was better controlled by beneficial arthropods. Much of this success came from the use of insecticides late in the production season to reduce the number of boll weevil adults leaving the fields to enter diapause. These applications are made after Heliothis is no longer a potential problem. While there are unknowns in the interrelationships among pesti- cide applications, beneficial arthropods, and Heliothis species, it is certain that the eradication of the boll weevil would sharply reduce Heliothis damage in some areas. An OPM program would also reduce Heliothis damage, though perhaps not as effectively. Success- ful integrated pest management (IPM) programs also have demonstrated the ability to reduce Heliothis damage. Biological Consequences of Failure If a boll weevil eradication program was instituted and later abandoned short of eradication, what would be the biological conse- quences? It can be assumed that the OPM programs that preceded the eradication program would have resulted in very low densities. Hence, damage to the cotton crop from weevils would not be of immedi- ate concern, and it would take several years for the boll weevil to return to preprogram density levels. In addition, growers who had participated in the OPM program would use their acquired knowledge to control other cotton pests, which would have a significant impact on boll weevil survival. Thus, the boll weevil population would be kept below damaging levels for undetermined but significantly long periods of time. If the eradication program was not completed, a decision to declare it a failure would not be clear cut. No such judgement could be made until a fairly lengthy period of time had passed. Boll weevil populations in eradication zones l, 2, and 3 (see Figure 5.l), for example, may never be reduced to levels below detec- tion. If boll weevils were still detected in zone l or 2 some 5 or 6

72 e Cotton Belt divided acres of cotton. The 0 The area within the economic status in 4J CO id •H 0 0 fC o H • 4J o rH CO in o O M 0 (4_j in A f] O 0) u CO 0 ;>, id m Jj A tU to c H rl 0) S 'H in Q) H io o> 0 •H M 0 •H ^ •o a to 0 0 OJ ^5 t0 0) ,— , 4J ^ O t0 CO CO u r-t 0 4J ifl o rH • co O H m C T) ^ c Q C rH •H <0 •H A o *••* X! N r-H O CO CO o •H 0) •H "3 4-' 9 c* jj U ;> H c 4J 4J <i^ •H 0 0 iO Q) CO 4J 0 C ^4 . 0) CO M •H 0) co y^ 0 CO r-H c OH u 45 ^ rH o Q) •H » iH a 0 •H M e C U £ 4J O 0 o •rl <0 0 c O* 4J 0 C O t0 c x; 0 •H u 0 o •H p( r-H 0 rl 4-1 4-1 O O 4J 10 c *o rti 10 o 4-1 •rH g 0 c u W t3 f£J •rH C c tj^ CO £ 0 rrt T3 o id U 4-1 $ £ •H 0 •o 4J 43 io 4-1 Vj 0 O •§ a a, tU M u CO A 4J u •H U-l 4J Lj CO tn •H 'H O 0 •H 1 E 4-1 C 1 O CO CO •H 0 tn C •H 0 x; H C ^ 0 TJ c 10 •rl Q u o •H •rH rH 0 u C N ^ 4J ^1 O C •o 4-1 O <0 0 •H 4J 0) 4-J 4-J 0 4J rH in O C M 0 O O iw u •H <0 a in E tn § D

73 years after the eradication program began, the program would most likely revert to an optimum pest management program. This form of failure would produce the least disruption in the cotton ecosystem and be the least expensive. The amount of insecticides used in retreating areas where boll weevils are found will increase over the original amount planned only slightly and over a relatively short time interval (5 to 6 years). A second possibility would be a reduction of the boll weevil population to such low levels that detection of small, incipient populations might require 8 or more years. If this occurred, the eradication program would have moved to the final zone before popula- tions reached detectable levels in the first 2 zones. Retreatment of these zones with insecticides could become a significant factor, and there could be major disruptions of IPM programs by greatly decreasing the effectiveness of natural enemies. Subjecting low-level boll weevil populations to insecticides could accelerate the development of natural resistance in the boll weevil. Failure of this type would be the most costly, both economically and environmentally. An evalu- ation of this type of failure might require more than a decade, and the failure might become a major political issue. A third form of failure could be eradication of the boll weevil from several zones and survival of a persistent population in others. This could be the result of a lack of cooperation between agencies and cotton growers, or it could have an unforeseen biological basis. In either case, the persistent population would be subject to intense insecticide applications over a limited geographic area. Secondary outbreaks could result, disrupting existing IPM programs. This could result in considerable economic loss to growers in that limited area. Failure would not be instantly recognized in any of these situa- tions, but the net result would be to greatly increase the use of insecticides, accelerate pest resistance to insecticides throughout the Cotton Belt, reduce the role of natural enemies, and increase the probability of secondary pest outbreaks. Biological Consequences of Success Initially, a beltwide boll weevil eradication program would have a large and beneficial effect not only because a key pest of cotton had been eliminated but also because cotton growers would become more knowledgeable about integrated pest management methods. The NRC Committee points out here some of the other possible biological consequences of a successful eradication program. As a result of boll weevil eradication the management of cotton production would change. Early season insecticide treatments would be greatly reduced, and this would probably have positive effects on populations of beneficial organisms. On the other hand, insect pests that had been held in check by insecticide treatments aimed at boll weevils might increase to the point where early season sprays again were required for their control. This would eliminate some of the economic and environmental benefits that justify an attempt at boll

74 weevil eradication. The research needed to evaluate this problem has not been done, and it would require the absence of insecticides for a period of several years to determine the response of other insect pests and beneficial arthropods. Since eradication of the boll weevil would reduce the need for insecticide applications, secondary pest outbreaks would be reduced because insecticides would no longer be destroying beneficial arthro- pods. But there is no information on other key pests that may be kept under control by the insecticides used in current boll weevil control programs across the belt. Another possible consequence of a successful eradication program would be a change from early maturing varieties of cotton to higher yielding, late-maturing varieties. This would tend to make cotton more susceptible in years favorable to the development of other pests and more vulnerable to reinvasion by the boll weevil. Probably the most serious biological consequence of successful eradication would be the significant increase in cotton acreage on land ill-suited for sustained cotton production, even though total cotton production would be expected to remain the same. Much of this land was taken out of cotton production as a result of severe boll weevil problems and associated pest control costs. This marginal land would be subjected to increased erosion and soil degradation if replanted in cotton. It is important to evaluate the biological and environmental consequences of both success and failure. Not all of the effects of failure would be negative, nor would all the effects of success be positive. What seems to be clear is that the BWE trial provided very little information for judging these effects. ECONOMIC CONSIDERATIONS The assignment of the USDA's economic evaluation team was to estimate the market consequences of the successful beltwide implemen- tation of each of the proposed boll weevil eradication and control strategies. Cotton yield increases and the cost reductions expected for specific locations were needed to make these estimates. USDA systematically accumulated the opinions of local experts through a Delphi approach, and the opinions of several experts from each area were averaged to identify the probable insecticide-use and lint-yield changes (Economics and Statistics Service l98ld) . The USDA economic evaluation team is to be commended for its creative efforts to obtain beltwide cotton production data, project the future, and develop market benefits and redistribution effects of successful implementation. But, the USDA team was less thorough and creative in developing data on the public cost and probability of success of each program. Based on the collection and averaging of divergent opinions of practitioners, the costs were less than the benefits. The future, however, is unknown and experts do not agree about either the cost or the benefit of public management or eradica- tion of the boll weevil. The market benefits estimated by the USDA

75 team assume successful implementation, but the estimated public costs do not guarantee beltwide success in either management or eradication. Several experts contributing to the Delphi survey indicated concern about a possible bias toward the success and feasibility of public programs. For example, extension service personnel believe that additional extension service educational activities and techni- cal assistance would result in improved cotton insect management and higher yields (Economics and Statistics Service l98lb). The USDA report admits that there is a degree of uncertainty in the Delphi estimates of control costs and yields, but no uncertainty is expressed about the cost estimates of beltwide OPM or BWE programs (Economics and Statistics Service l98lb). The considered opinion of the NRC Committee is that there is considerable uncertainty in both the benefit and cost estimates, and that this uncertainty is suffi- ciently high to preclude their being used as a basis for deciding between programs. The Committee believes that producers in some regions might benefit considerably in economic terms from public boll weevil management or eradication but that the nation's total agricul- tural production would not change very much. Consumer Benefits The USDA economic evaluation report concludes that a successful eradication or control program would lower the cost of producing a pound of cotton in some areas and thus encourage more cotton to be produced. The result of this increased efficiency and production of cotton would be to reduce the price of cotton and other agricultural products to consumers in the U.S. and abroad. This conclusion is not surprising, since consumers generally have benefited from past im- provements in agricultural productivity. In dollar terms, the USDA report (Economics and Statistics Service l98lb) presents consumer benefits for five different types of program (OPM-NI, OPM-PI, OPM-I, CIC-BWE, and OPM-NI-BWE) over the long term—these are the total present dollar values of all future years' benefits. A discount rate of 7.l25 percent was used. About 40 percent of these estimated consumer benefits are from a reduction in the price of cotton of between 2.2 and 3.6 percent (or l.7 to 2.7 cents per pound), and the remaining 60 percent are from the expected reduction in the consumer costs of the cotton seeds and crops that comprise the U.S. feed-grain complex—namely, soybeans, corn, cotton seeds, soybean meal, and so on. The total consumer benefits are estimated to range from $4.l7 billion for the CIC-BWE program to $6.46 billion for an OPM-NI-BWE program (see Table 5.2). The net benefits of each of the five programs are the above benefits to consumers minus the cost of the program and minus a loss in net income to cotton and other agricultural producers. The net market benefits range from $2.44 to $3.89 billion (see Table 5.2). The estimated program costs to be paid by the government are small relative to the net market benefits. The USDA report states that consumers could afford to compensate agricultural producers for

76 TABLE 5.2 Present Values of Benefits and Costs for Alternative Boll Weevil Management Programs —' Group or Item C hanges i n Present Values ^ f OPM-NI OPM-PI OPM-I CIC-BWE OPM-NI -BWE ________ 4.50 — Pt'i Tl'i f^rt Dollars —-• 4.l7 1 1 Consumer benefits £/ 4.58 5.l6 6.46 Net income to cotton -.85 -.84 -.60 -.42 -.96 producers Net income to other -l.l0 -l.09 -l.04 -.84 -l.37 producers d/ Program costs paid by the government — .06 .l2 .44 .l6 .24 Net market 2.57 2.45 3.07 2.75 3.89 benefits f/ B/C ratio 2/ 44:l 2l:l 8:l l8:l l7:l - I 1 5/ Net benefits and B/C ratios are based on unrounded data. Represents changes in present values of benefits and costs as compared with a baseline representing current insect control. b/ Future benefits and costs in l979 dollars, discounted at a 7.l25 percent rate in perpetuity. £/ Consumers include all market participants beyond the farm gate, including processors, mills and final consumers. £./ Includes producers of soybeans, corn for grains, grain sorghum and small grains. £/ Producers were assumed to pay 50 percent of eradication program costs, exclusive of capital costs and follow-up monitoring. Producer shares of program costs are reflected in returns to cotton production. £/ Net market benefits equal the sum of above consumer and producer benefits less program costs paid by the government. Generally considered best criterion if there are no budget constraints. 3/ B/C ratios are calculated as the sum of consumer and producer benefits divided by public program costs. Generally considered best criterion if there are budget constraints. SOURCE: Economics and Statistics Service (l98la)

77 any losses they suffered in net income as well as pay the full public costs of the program and still come out ahead. The NRC Committee, however, believes that the probability that consumers would benefit to such a degree from a boll weevil control or eradication program, which would cost so little, is extremely small. According to the NRC Committee's estimates, the reduction in insect control costs plus the value of the increased cotton yield on historical cotton acreage would result in a productivity gain of only about $l25 million per year. Since the United States already pro- duces a cotton crop each year that is worth about $5 billion, such a productivity gain would amount to only a 2.5 percent gain over l0 years, or 0.25 percent a year. An expected productivity gain of this small magnitude could have little effect. U.S. agriculture as a whole produces about $l40 billion in agricultural commodities each year. It seems unlikely to the NRC Committee that such a relatively small improvement in cotton productivity would depress the net in- comes of other producers by $l billion as the USDA report predicts (see Table 5.2) . For example, USDA estimates that the cottonseed crop also would rise by 2.8 percent; this would mean an increase in the cottonseed crop each year of about l30,000 tons of seeds. Total U.S. production of oil seeds from all types of crops, however, amounts to about 6l million tons a year. Thus, the increase in the cottonseed crop would amount to only 0.2 percent of the annual oil seed crop. The USDA economic evaluation report estimates that such an increase would reduce the value of the cottonseed crop from the present $580 million a year to somewhat less than $500 million. A more realistic view, in the NRC Committee's opinion, is that the price of cotton seeds would decline by a smaller percentage than the percentage of increased production. The result would then be a slight increase in the value of the cottonseed crop. The USDA economic evaluation report correctly recognizes that any control or eradication program for cotton boll weevil would have positive impacts on net incomes of cotton producers in areas inhab- ited by the boll weevil but negative effects on the other areas and producers of other commodities. Such redistribution consequences are important to individuals significantly affected. The NRC Committee believes the pattern of the redistribution estimated by the USDA team is correct, but believes that the extent of the redistribution has been overestimated by USDA. The USDA's economic evaluation does not report the consequences of the sequential reduction in production costs that would occur as the eradication program moved across the boll weevil-infested areas of the Cotton Belt. Producers in the initially eradicated zones would receive the advantages of production cost reductions, yield increases, and reduced risk without the disadvantage of reductions in cotton prices. Conversely, producers in the last zones to be eradi- cated would suffer the disadvantage of price reductions without the advantages. This has already begun to occur in the eradication trial area of North Carolina, where there has been a 75 percent increase in the acres planted in cotton since l979.

78 Economic Consequences of Program Failure There are several reasons why a beltwide eradication or control program might not be completed after it had been started. These include a lack of the necessary funding from the federal government, state governments, or the growers themselves, a failure to obtain the necessary regulatory laws through legislative action or by means of a grower referendum, or a failure of the program itself to achieve eradication in its early stages. The probability that the program might come to a premature end would be increased to the extent that its economic costs were understated. A premature end to the program would cause economic harm to all of the producers whose farms had not yet been reached by the program. These producers would have suffered the economic damage of a decline in cotton prices without ever benefitting from the reductions in costs postulated as results of the program. Evaluation of Program Costs In l973 the Stanford Research Institute (l973) conducted a study designed to estimate the costs of beltwide eradication of the boll weevil. That study, which was based on the costs and results of the l97l to l973 PBWEE experiment, projected the high cost of a beltwide eradication effort at $2.46 billion and the low cost at $l.ll billion. These projections, corrected to take account of the inflation that has occurred since then, would run from $2.24 to $4.9 billion. In contrast, the USDA economic evaluation report implies that an initial appropriation of $240 million, invested at an interest rate equal to future inflation rates plus 7.l25 percent, would be suffi- cient to pay all of the government costs of an eradication program (see Table 5.2). The NRC Committee believes, however, that the probability of successful beltwide eradication from an appropriation of only $240 million would be extremely small, for at least three reasons: • The initial required capital and facility outlays are not taken into account in the estimate of program costs. • The estimated cost of future eradication operations appears to be only a fraction of the operating costs actually experi- enced in the BWE trial in North Carolina. • The eradication trial in North Carolina—conducted after two severe winters had reduced the boll weevil population—lasted three years but beltwide eradication costs include only two years of program costs. A larger probability of success would necessarily be associated with a larger appropriation for the beltwide OPM or BWE program. To achieve a very high probability of success with the BWE program—for

79 example, 95 or 99 percent eradication within l0 years for the entire Cotton Belt—it would be necessary to make an appropriation of per- haps as much as $5 and $l0 billion, and to give government authori- ties the power to compel near l00 percent participation by cotton growers. The USDA report should have included cost-benefit ratios for different levels of costs for both OPM and BWE. Higher costs and smaller benefits would be associated with less favorable weather, less cooperation from growers and state governments, and delays or repetition in executing the program. Because the economic evaluation report does not include alternative costs and alternative benefit- cost ratios for each program, an informed and intelligent decision about which program to select, if any, cannot yet be made. The NRC Committee feels that the cost estimates shown in the USDA economic evaluation report for the three types of OPM programs are underestimated. For example, estimated costs for scouting and aerial application of insecticides are unrealistic in reference to current operational practices. The NRC Committee therefore urges that an accurate estimate of program costs be developed by an independent agency prior to any USDA request for public funding of a beltwide eradication or management program. ENVIRONMENTAL CONSIDERATIONS The USDA environmental evaluation team report (APHIS l98la) was prepared by members of the Environmental Evaluation Staff of APHIS and a subcontractor, Ketron, Inc., who provided an analysis of the trial results (Miller and Carpenter l979, Carpenter and Miller l98la, l98lb). Additional unpublished data were provided by APHIS on insec- ticide residues in North Carolina and Mississippi during the first two years (l978 and l979) and later for the third year of the trials. These data consisted of samples of soil, vegetation, water, sediment, insects, mammals, birds, and fish collected under APHIS supervision. Samples were analyzed for l4 insecticides at the USDA Gulfport, Mississippi laboratory (APHIS l98la). The contractor used the BOLL-l Model created by Ketron Inc. (Arlington, VA), the principles of which have been outlined by Carpenter and Miller (l98la) and consists of seven modules, to address different aspects of environmental impact combined and nor- malized to produce an overall Q index. The seven modules were: • off-site pesticide drift, • human ingestion, / • research conflicts, • fish farms and hatcheries, • endangered and threatened species,

80 • wildlife, and • aquatics. The combined Q index involved weights for impacts determined by a Delphi survey of scientists (Carpenter and Miller l98la) and thresh- old values representing the level of hazard not to be exceeded if the impact is to be judged acceptable. Monitoring the Environmental Effects of OPM, CIC, and BWE Trials Prior to the time the trials began, the major value of monitor- ing their environmental effects appeared to be the generation of information that could be extrapolated to beltwide programs. Second- arily, techniques for assessing environmental effects could be tested in the trials with a view toward evaluating them for a beltwide effort. Sampling, analytical procedure, the applicability of simula- tion models, and problems with extrapolation might be so evaluated. A beltwide boll weevil eradication or control program would require a thorough analysis of potential environmental risks. Such an analysis would presumably be similar to an environmental impact statement (EIS)—either as a legislative requirement or at least in the spirit of the National Environmental Policy Act (NEPA) of l969. Recent guidelines issued by the Council on Environmental Quality for environmental impact statements require that an initial public statement be made defining federal objectives, raising significant environmental issues (both risks and benefits), and examining alternatives. Therefore, any evaluation of a beltwide boll weevil management program should examine all alternatives—including no management, continuation of current practices, implementation of new control strategies, and eradication—in light of the probability of achieving the desired objectives, the economic benefits versus costs, and the environmental risks and impacts. Generic Environmental Issues Management Concerns. There were several inconsistencies in the Delphi process from which the insecticide application estimates were derived. The estimates were strongly influenced by persons familiar with CIC practices who had limited knowledge of new management tech- niques being developed in the trials. Similarly, it is unclear how the detailed management program from the OPM and BWE trials can be extrapolated to take account of the diversity of regional pest and crop relationships throughout the Cotton Belt. Environmental Constraints on Boll Weevil Management. A number of environmental factors, both controllable and uncontrollable, would influence the effectiveness of the various proposed programs. The presence and density of alternate plant hosts, and the conditions for

8l reinttoduction of weevils, would affect not only the efficacy of eradication but also the extent and location of buffer zones. The availability and costs of water for irrigation, fertilizers to aug- ment regional differences in soil fertility, and methods of managing other cotton pests could affect the future distribution of cotton cultivation, with or without boll weevil eradication. These factors should have been considered in any beltwide plan. Short-Term vs. Long-Term Impacts on Environmental Quality. An intensive beltwide eradication program would involve a trade-off between the ecological risks of short-term increases in insecticide residues and the long-term benefits of an overall reduction in insec- ticide use once eradication was achieved. The eradication effort might fail to reduce insecticide use in the long run, however, either through failure to eradicate the boll weevil or, if eradication was achieved, through the continued use of insecticides to control other pests. Catastrophic Effects on Biota. Despite statements to the con- trary (Economics and Statistics Service l98la), it remains to be demonstrated that species capable of withstanding current cotton insect practices would not be seriously affected by an OPM or BWE program. During the initial stages of a BWE program, heavy and repeated applications of general biocides (such as methyl parathion) would be applied to a major part of the agricultural environment. These might cause irrevocable damages to life forms that would not recover at subsequent lower insecticide levels. Examples include honeybees, other pollinators, biological control agents, endangered or threatened species without the resiliency for population recovery, and coastal and estuarine shellfish, whose extremely low tolerance for diflubenzuron could result in substantial kills. The occupa- tional exposures of workers and the potential effects of increased insecticide use on the health of human populations in the region also deserve careful scrutiny. Effects on Land Use. Boll weevil eradication could influence land use, although it is unclear whether the current patterns of cotton acreage would change. It is also unclear where, and if, cotton would be competitive with other agricultural land uses, or if, with increased cotton yields, the acreage planted would be increased or decreased, or if fallow lands and those in plantation sylviculture would be converted to cotton production. Other external factors, such as water availability and energy costs, could conceivably direct future demands on agricultural productivity to areas formerly unim- portant agriculturally. The potential environmental effects of intensified southern agriculture in general, and cotton agriculture in particular, should therefore be examined, issues that need to be addressed include increased soil erosion, the effect of concomitant increases in suspended solids on surface water quality, and degrada- tion of the soil fertility of soils in the South that are still recovering from previous intensive agricultural use. None of the aforementioned issues have been adequately dealt with in the initial environmental evaluations of beltwide boll weevil management alternatives by the USDA team. Before the adoption of any management strategy, particularly beltwide eradication, these issues

82 should be scientifically examined. Outlined in the following sec- tions are the environmental factors, types of data, and sources of information that should be utilized in such an assessment. Environmental Toxicology and Health One of the major externalities resulting from the massive em- ployment of insecticides for cotton insect control is their effect on non-target organisms and on overall environmental quality. Although there are at present at least 36 insecticides registered for cotton insect control, a relatively small number are predominant (see Table 2.2). Before l946, calcium arsenate comprised more than 90 percent of the total amount of insecticide applied. From l946-l955, DDT and toxaphene comprised at least 90 percent of the total applied. By l964 (the time of the first reasonably accurate survey) toxaphene comprised 34 percent, DDT 30 percent, and methyl parathion ll percent of the total. By l97l, toxaphene was 38 percent; DDT, 18 percent; and methyl parathion, 3l percent. By l976, toxaphene comprised 44 percent; methyl parathion, 3l percent; and EPN, l0 percent. The mix changed rapidly because of pest resistance, local pest problems, withdrawal of compounds by the manufacturer (e.g., chlordimeform in l976), ana legal restrictions on the use of DDT in l973, aldrin and dieldrin in l974, heptachlor and chlordane in l976, and endrin in l979 (NRC l975) (see Table 2.2). It is estimated that a total of about 2.3 x l09 ibs. of insec- ticide ( >200 lbs. per acre) have been applied to U.S. cotton soil— largely calcium arsenate, 850 x l06 (total >80 lbs. per acre); DDT, 500 x l06 ibs. (>50 lbs. per acre); toxaphene, 600 million lbs. (>60 lbs. per acre) ; and methyl parathion, 300 x l0*> lbs. (>30 Ibs. per acre). The overall environmental effects have varied but have included widespread soil contamination with persistent residues of calcium arsenate, DDT, and toxaphene; large kills of fish; wide- spread mortality and decreased reproduction of terrestrial wildlife; substantial mortality in honeybee populations; decimation of benefi- cial parasites and predators; and ubiquitous contamination of the bodies of fish, mammals, birds, and human beings. Unfortunately, systematic data to quantify these effects are scarce, but some idea of the prevalence of insecticide residues in the Cotton Belt can be gained from data provided by the national soil and air monitoring programs of the Environmental Protection Agency shown in Tables 5.3 and 5.4. Heavy use of calcium arsenate in cotton-growing areas has left residues that have persisted for 25 years or more. Compared with the national averages for cropland soils, residues of DDT and its breakdown products (referred to as DDT-Total or DDT-T) and of toxaphene occur at substantially higher levels in the cotton-growing states (Table 5.3). The air-monitoring data from suburban locations in three cotton- belt cities (Montgomery, AL; Little Rock, AR; and Monroe, LA) show substantially higher airborne residues of DDT-T, toxaphene, methyl parathion, and endrin than are present in cities monitored outside the cotton-growing area (Table 5.4).

83 V f> »H ro iH r* (0 ^J V M O O O VO O O O O • 1 A i oo r- ^ IH o fM iH O O i-l ^ CO ro iH O O O % A 0) __1 1 CJ 00 ro 00 iH 00 r-l in O O CM VO O O O O 'c A •H O VO rH r-t 00 £X| r^ o o o r» o w i 8 • • • • • 5 in CM o o o in c* <N ^ r- •H ch o o m 1 3 ri a d o o .5 A iH VO iH CN O i n VO 0^ O O VO 01 u CN O O O O A 4J id J3 m CO 2 5 in r~ <N IH vo t-l rr O O r-t •0 m rH o o o 'u •H 00 VD 1^ 0^ I a in r- o O .5 VD O 1 O rH r-, oo t^ <N rH r* 0^ <M CTi VD O O (N •-i O 0* • • • • • ^•^ s (^J 00 O O O 0 rH 3 10 •H i-l m iH iH CT< •-t .H O O VD 4J £> 1j • • • • • 0) 1^ VO rH O O O c A CQ flj E 5 • x 0) 01 •H n •o « •H 0) in u C C •H U -H 01 •• u 4J •>H h C f. U M U O (4 *O *H Q< y ffl 0) 0) 1 rH M fl PH B C0 W E-i 0) T3 X fc Q -H C 0 1 H < Q Q W EH

84 0) c O 0) H O vO Q) c ^i « ro 0^ ft £ o^ 2 2 2 r- id f'J'HOQQQQQQ p| o> X rH ^ CN A Z Z Z Z Z id p^ i o H X g 1 c £ ^ i O .2 3 ) 3 5 1 id id ) M Cf> j id rH^iHf^CO CD (T» id o « fi S a J I* •WvDfNnr-izzfNO u I H iH 4J ^J J^ 0) O Q S z S 1 0, 1 y i u c Q * -o o « J> -H c •H vO f) <"> JS Sro CM : u o •H h * * * Oi Q C% Q Q (3 ^J 0002ZZZZZ id 1-1 4-> 4 U in Q O r-l 0) u IH 2 z •H J S s •g 3 H <d c c i •H M *n o cNinococMOtnHvo c CN VO GO \f» r4 *f to : u f-4 iHf-liHOOiHOrHiH N r> o o Q U, u 4 0) id CN 55 c •H •H id i Q Q -P 0) CA -H C GO CN CO U •H 0, " U i g : r- g pH •H ~ ^H O § . . ^t O 4 r» f^ E-i 4 H 14-1 > U O • I : « U g 1 •H 1 M •H n co o <4-l J r» Q vfl in ^ r» 1 CA 1 JH r-l H in vD * ^ % ^ i u 04 ft ) 0) : ,P f*» * 3 « Si J 4-> V CA H M » & * T3 id :n -O a m fl) U 4 4-1 C f*» flj ^ Qj u • l> -H O > C -H 2 £ < ^ §2 w i m M 4-1 13 I X .H . H 6 * v b H 1-1 « V C • * J* H O U «• 1 •o 2 i ! u >, o < 2 wo) D M 6 J k S t? IJ 0) H . * I g 1 O tt) 0) 4-> ^ 3 «ld>rH -H 0) C C > iH 0) O C id *O 4J 0) > id id || p^ ! 1 0) T> W 0) H -H id £ e 8 idO'HO^oid'HOOO 3 x at * Z CA

85 Monitoring of living organisms reflects the generally higher pollution of the Cotton Belt environment by insecticides, although there are few comprehensive surveys. A comparison of residues of persistent insecticides in the wing tissues of woodcock (McLane et al. l978) showed much higher residues of DDT, dieldrin, and hepta- chlor epoxide in birds from Louisiana and the "tri-states" (Georgia, North and South Carolina) than from Maine or Michigan (Table 5.5). The national monitoring program for persistent pesticides in human adipose tissues shows that DDT residues among inhabitants of the Cotton Belt are substantially higher than those of other states (Table 5.6). The effect of specific insecticides upon environmental quality has been severe. Toxaphene, or chlorinated camphene, has had the widest use of all the persistent organochlorine insecticides on the cotton crop. Composed of at least l75 individual chlorinated ter- penes, of which 2,25-endo-6-exo-8,9,9,l0-octachlorobornane is the most active ingredient, toxaphene is water-soluble to about 0.40 ppm and has an average octanol/water partition of 825 (Sanborn et al. l976) . Toxaphene, on the average, persists in soil for 3 to l0 years and up to 6 years in water; it is bioconcentrated up to l00,000-fold in fish (NRC l975) . As shown in Tables 5.3 and 5.4, toxaphene resi- dues in the Cotton Belt are high and ubiquitous. Toxaphene is highly toxic on an acute basis. The rat oral LD5Q is 69 mg per kg; guinea pig, l5; mallard, 7l; pheasant, 40; bobwhite, 85; and sharp-tailed grouse, l0-20. Toxaphene is extremely toxic to fish, with LC5Q values for rainbow trout 0.0028, bluegill 0.0035, and black bullhead 0.005 ppm. For aquatic invertebrates, LCso values range from 0.006-0.l80 ppm (Pimentel l97l). Fish in water exposed to toxaphene levels as low as 0.0005 ppm suffer from "broken back" syndrome, a crippling collagen deformity (Mehrle and Mayer l977). Toxaphene has been demonstrated in several bioassays, including the National Cancer Institute bioassay, to be highly carci- nogenic in rats and mice (Reuber l979). Methyl parathion, or o,o-dimethyl o-p_-nitrophenyl phosphoro- thionate, is the organophosphorus insecticide most widely used on the cotton crop. Unlike toxaphene, methyl parathion is substantially biodegradable, although soil persistence up to 5 years and water persistence up to 2 years have been recorded (NRC l975). Methyl parathion has a water solubility of about 20 ppm and an octanol/water partition of about 800. Methyl parathion is not bioconcentrated to any marked degree. It is, however, an extremely toxic insecticide on an acute basis: the oral LD50 for rats is l4-24 mg per kg; for mallards, l0.0; and for pheasants, 8.2. The LC50 values for fish are 2.75 ppm for rainbow trout and 5.72 ppm for bluegill. For aquatic invertebrates, LC50 values range from 0.005 to 0.070 ppm (Piraentel l97l). The widespread use of methyl parathion is a constant threat to the health of farm workers because of its high toxicity when inhaled or contacted by skin. In cotton-growing areas in Texas, 25 percent of the workers loading spray planes reported acutely toxic effects.

86 TABLE 5.5 Average insecticide residue levels in woodcock wing muscle tissue (ppo), 1971-1972. State DDT DDE Dieldrin Heptachlor epoxide Tri-state area 6.82 26.03 1.64 0.51 (GA, NC, and SC) LA 2.97 9.80 2.46 0.69 ME 1.19 4.07 0.16 ND MI 0.92 3.16 0.20 ND ND = none detected. SOURCE: McLane et al. (1978) The extent of methyl parathion poisoning among field workers, flag- men, and residents—in particular, children—is unknown because of inadequate reporting practices. The use in El Salvador in l972 of 4,800,000 lbs. of methyl parathion and its even more dangerous rela- tive parathion resulted in 2,86l reported cases of poisoning and 30 deaths. About 2,560,000 pounds of active parathion ingredients were applied to U.S. cotton in l97l (see Table 2.2). The World Health Organization estimates that about 500,000 accidental poisonings and about 20,000 fatalities occur worldwide each year, largely from the use of methyl parathion and parathion (Copplestone l977). Other insecticides used to control cotton insect pests are also hazardous to environmental quality and human health. Endrin, or l,2,3,4,l0,l0-hexachloro-6,7-epoxy-l,4,4a,5,6,7,8,8a-octahydro-l, 4-endo,endo-5-8,-dimethanonaphthelene, is the most toxic of the widely used organochlorines. The rat oral LD5Q is l.5 to l7.8 mg per kg; mallard, 5.6; pheasant, l.8. Endrin is the most toxic insec- ticide to fish, with an LC50 value for rainbow trout of 0.00l8 ppm, and for blue gills, 0.00035 ppm; its UC$Q values for aquatic inver- tebrates range from 0.002 to 0.020 ppm (Pimentel l97l). Endrin has an octanol/water partition of l,600 and a water solubility of 0.06 ppm. It bioconcentrates to at least 3,000-fold. Endrin has a soil half-life estimated at 4 to 8 years. Endrin has been responsible for widespread fish kills and damage to vertebrate and invertebrate

87 TABLE 5.6 Average insecticide residue levels in ppm in human adipose tissues, 1968. State Insecticide DD-T White SOURCE: U. S. DHEW (l969) Black Dieldrin White Black AL 5.92 11.36 0.0l 0.00 AR 13.23 28.86 0.2l 0.l0 GA 9.68 11.70 0.18 0.l9 LA 8.55 14.75 0.10 0.06 NC 7.73 13.2l 0.11 0.ll TE 9.28 14.14 0.l7 0.08 22-state average 6.32 12.06 0.12 0.l4 wildlife (Pimentel l97l). Approximately l,068,000 lbs. (active ingredient) of endrin were applied to U.S. cotton in l97l (see Table 2.2) . The organophosphorus insecticide EPN, or o-ethyl o-p_-nitrophenyl phenylphosphonothioate, is also extremely hazardous. It has a rat oral LD50 of 7.7 to 36 mg per kg; mallard, 3.l; pheasant, 53; partridge, l4. Few studies have been made of the environmental effects of EPN, but it is known to be a delayed neurotoxin that can produce irreversible paralysis in chickens at very low levels of acute or chronic ingestion. About 6,l40,000 lbs. (active ingredient) of EPN were applied to U.S. cotton in l976. Data provided by the USDA environmental evaluation team from Delphi estimates of the insecticides likely to be used and their application rates in a beltwide eradication program, as well as data on the insecticides used in the OPM and BWE trials, indicate that l5 chemical formulations (and diflubenzuron, which is not included in

88 the table) are likely to be employed (Table 5.7). Four insecti- cides—dimethoate, endrin, fenvalerate, and sulprofos—were used in the OPM trial but not in the BWE trial (Table 5.8). All formulations were used in at least two of the three trial years, except for ace- phate, azinophosmethyl, and chlorpyrifos in the OPM trial. Of these l6 insecticides, only 7 (chlorpyrifos, diflubenzuron, EPN, fenvaler- ate, methyl parathion, permethrin, and toxaphene) were monitored by APHIS in the environment surrounding the North Carolina and Missis- sippi trial areas. During the three years of monitoring by USDA, only 3 insecticides used in the trials were detected (Carpenter and Miller l98la). Methyl parathion (mean cone. 0.0l9-0.06l ppm, max. cone. 0.477-0.830 ppm) was detected in 6 percent of 3l5 samples of avian tissue, and chlor- pyrifos (mean cone. 0.004-0.506 ppm, max. cone. 0.237-0.300 ppm) in l percent of the samples. No residues were detected in 60 water sam- ples and 70 sediment samples during the trial years. No fish or aquatic biota were analyzed. It is difficult to understand why there were not more signs of residues, why only a small number of the insecticides were detected, and why parathion and malathion were detected even though they were not reported to have been used in the trials. The absence of positive signs of residues was certainly due, in part, to the fact that not all l6 insecticides were analyzed for. The environmental sampling design and analytical techniques used in the monitoring were not explained. At best, however, the data are entirely inadequate for extrapolation to a beltwide level. The known toxicities of various insecticides to non-target biota are shown in Tables 5.9, 5.l0, and 5.ll. The few data reported from the monitoring program for birds and mammals are well below LD5Q values. Aquatic organisms, however, are much more sensitive to nearly all of the insecticides used. The extensive literature on insecticide-related fish kills makes it very hard to understand why aquatic biota were not monitored. The measurement of insecticide residues in soils should have been initiated before any additional insecticides were applied and on a continuing basis afterwards. This would have provided a pretrial background measurement of insecticide residues in agricultural soils and measurements of the rates of accumulation. These accumulation rates, together with information on rates of applications, would have been valuable for making beltwide extrapolations. Moreover, the soil analysis appears to have proceeded independently of information about actual applications of insecticides. For example, an attachment to the USDA environmental evaluation report (Carpenter and Miller l98la) lists l5 insecticides applied to the trial area in Panola County between l978 and l980, however, the analysis of pesticide residues did not include most of those insecticides that were used in the trial area. Organisms in aquatic ecosystems are notoriously sensitive to insecticide damage. Field measurements in small calibrated drainage systems would have made it possible to estimate the amounts of insec- ticides reaching aquatic ecosystems and their subsequent dilution, degradation, concentration in sediment and biota, and effects on

89 TABLE 5.7 Cotton insecticide data for 0PM fields in Panola County, Mississippi. Chemical and application rate (lbs/acre) Average number of pounds per acre in the OPM applied area 1978 1979 l980 Acephate l.30-2.00 0.04 - Az inopho sme t hyl 0.l6-0.25 - 0.0l Chloridimeform 0.05 0.03 0.l3-0.25 Chlorpyrifos 0.33-0.50 0.01 — Dicrotophos 0.l0-0.25 0.09 0.35 0.07 Dimethoate 0.02 0.35 0.07 0.l0-0.20 Endrin 0.0l <0.0 0.27 EPN 0.28 0.38 0.22 0.50-0.75 Fenvalerate 0.03 0.l2 0.ll 0.05-0.20 Met homy 1 0.30-0.45 0.22 0.33 <0.0 Methyl parathion 0.25-l.50 0.28 0.82 2.29 Monocro topho s 0.20-l.00 <0.0l 0.0l 0.02 Permethrin 0.l3 0.03 0.03 0.l0-0.20 Toxaphene l.50-2.00 0.05 0.52 - Sulprofos 0.50-l.50 0.07 0.0l SOURCE: Carpenter and Miller (l98la)

90 TABLE 5.8 Cotton insecticide data for CIC-MS sample fields in Pontotoc County, Mississippi. Chemical and application rate (lbs/acre) Average per number of pounds applied acre in the OPM area l978 l979 l980 Acephate l.30-2.00 < 0.0l < 0.0l Azinophosmethyl 0.l6-0.25 0.08 - 0.06 Chlori dime form 0.l3-0.25 0.03 - 0.01 Chlorpyrifos 0.33-0.50 0.0l 0.l4 - Dicrotophos 0.l0-0.25 0.ll 0.0l - EPN 0.50-0.75 0.06 - 0.ll Methomyl 0.30-0.45 - 0.07 0.l2 Methyl parathion 0.25-l.50 0.02 0.7l 0.69 Monocrotophos 0.20-l.0 0.l0 0.0l - Permethrin 0.l0-0.20 0.02 0.0l 0.06 Toxaphene l.50-2.00 0.0l 0.72 0.l3 SOURCE: Carpenter and Miller (l98la)

91 TABLE 5.9 Avian toxicity data for selected insecticides. Pesticide Organism LD50a (mg/kg) Carhamate Lannate Insect growth regulator Mallard duck Quail 15.9 l5.0 Di f lubenzuron Mallard duck Quail >2000 >5000 Organochlorine Toxaphene Mallard duck Ring-necked pheasant 70.7 Organophosphate Azinophosmethyl 40.0 Mallard duck Ring -necked pheasant 136 74.9 Chlorpyrifos Mallard duck Quail 75.6 l6 Dicrotophos Mallard duck Ring-necked pheasant 4.24 3.2l Dinethoate Mallard duck Wild bird 4l.7 50.7 EPN Mallard duck Ring-necked pheasant 3.08 53.4 Malathion Mallard duck 1485 Methyl parathion Mallard duck Wild bird l0.0 50.7 Monocrotophos Mallard duck Ring-necked pheasant 4.76 2.83 Synthetic pyrethroid Permethrin Mallard duck >4640 Fenvalerate Mallard duck >9932 is defined as the lethal dose to 50 percent of the test population. References for the source of these data are given in Carpenter and Miller (l98la). SOURCE: After Carpenter and Miller (l98la).

92 TABLE 5.l0 Acute mammalian toxicity data for selected insecticides. Pesticide Organism LD5Oa (mg/kg) Carhamate Lannate Insect growth regulator Diflubenzuron Organochlorine Toxaphene Organopho sphate Azinophosmethyl Chlorpyrifos Dicrotophos Dimethoate EPN Malathion Methyl parathion Azodrin Synthetic pyrethroid Permethrin Fenvalerate Mule deer Rat Rat Mouse Mule deer Mouse Mouse Rat Mouse Rat Mouse Rat Mule deer Rat Mouse Rat Rat Mouse Rat Mouse Mule deer Rat Rat Rat ll-22 27 4640 4640 l39-240 ll2 7.l5 l3 l52 l45 11 16 >200 l52 42 8 l400 886 l2-l6 l8.5 25-50 2l 4l0 45l aLD50 is defined as the lethal dose to 50 percent of the test population. References for the source of these data are given in Carpenter and Miller (l98la). SOURCE: After Carpenter and Miller (l98la).

93 TABLE 5.11 Aquatic toxicity data for selected insecticides. Pesticide Test Organism LC50a (ppm) Car hamate Lannate 24-h Channel catfish 0.92 Insect growth regulator Di f lubenzuron 96-h Channel catfish 370 Di f lubenzuron 96-h Rainbow trout 240 Organochlorine Toxaphene 96-h Pinfish 0.0005 Toxaphene 96-h Fathead minnow 0.0l4 Organophosphate Guthion 96-h Brown trout 0.004 Guthion 96-h Catfish 3.29 Malathion 96-h Bluegill 0.103 Malathion 96-h Catfish 8.97 Methyl parathion 96-h Bluegill l.6 Methyl parathion 96-h Catfish 5.7l Methyl parathion 96-h Crayfish 0.003 Synthetic pyrethroid Permethrin 96-h Bass 0.0085 Permethrin 96-h Channel catfish 0.00ll Permethrin 96-h Crayfish 0.00062 Fenvalerate 24-h Rainbow trout 0.02l is defined as the concentration of insecticide in the water that is lethal to 50 percent of the test population. References for the source of these data are given in Carpenter and Miller (l98la). SOURCE: After Carpenter and Miller (l98la).

94 biota. Aquatic systems linked with the cotton fields should have been the major ecosystems sampled. The aquatic samples collected and analyzed were insufficient in terms of the variety of environmental components sampled and the numbers of replicates taken. Environmental Quality The monitoring of terrestrial biota could have included system- atic sampling of beneficial insects, pollinators such as honeybees, insectivorous birds, raptors, and small mammals. In this way the concentration of chemicals and their effects on sensitive organisms might have been documented. As it was, only vegetation was routinely sampled. The BOLL-l model allows various environmental factors to be included in an overall index, Q, for each insect control alternative. Of the seven modules, the Endangered and Threatened Species Module seems to have been particularly well evaluated (Carpenter and Miller l98la) . The index yielded a significant environmental impact value for the BWE trial in l979, but subsequently Q was set arbitrarily at zero for l980. The Offsite Drift Module is well explained, but no actual measurements of drift were made that would have allowed an evaluation of its assumptions. The role of wind in dispersal was not considered, nor was the presence of other crop sprayings. The Human Ingestion Module assumes a rather high drift of sprayed insecticides to non-target agricultural land (25 percent) but assumes that drift would be equally apportioned over non-target areas. This module might have been made insecticide-specific to account for differences in persistence, toxicity, bioaccumulation, and so forth. Different pathways of exposure should have been included (inhalation, water intake, fish and wildlife consumption), as well as the information on the current background level of insecticide intake for humans. The Wildlife Module used white-tailed deer as an indicator species, and hunter-kill records were used to evaluate changes in the trial areas. This was a dubious procedure to use, considering the size of the OPM area and the short-term nature of the records. Game birds (quail, mourning dove) might be more significant indicators than deer in parts of the trial area. In l980 the wildlife module was set at zero. The USDA environmental evaluation team judged the methodology to be unsuitable. The Research Conflicts Module and the Fish Farms and Hatcheries Module both yielded zero results. No research conflicts arose, and no fish hatcheries were located near the trial areas. Obviously, extrapolation to beltwide programs would change the results for these modules. The Aquatics Module appears to have failed because of lack of data. No field data appear to have been collected, nor were there adequate bodies of water for sampling biota near the fields. The Overall Index Module ultimately depended, therefore, on two indices with non-zero values, namely, Offsite Drift and Human Inges- tion. Other modules either did not apply to the trial areas (Re- search Conflicts, Endangered and Threatened Species, and Fish Farms)

95 or were not adequately measured (Wildlife, Aquatics). Comparisons of overall indices calculated for OPM, BWE, or CIC trials thus were based on minimal considerations, and extrapolation to beltwide pro- grams would not be valid. The overall indices differed greatly between l979 and l980, and the contractor's report (Carpenter and Miller l98la) and summaries by APHIS (l98la) and the Economics and Statistics Service (l98la) dif- fered in their results. Table 5.l2 compares the Q index for these two years for the two trial areas and the current insect control comparison areas (Carpenter and Miller l98la, Economics and Statis- tics Service l98la). The Q index for 0PM increased by a factor of five in the second year, while the Q index for BWE shrank by a factor of seven. The index thus classified BWE as the least desirable alternative in l979 and the most desirable in l980. The difference between years for OPM may have been due principally to a change in insecticide use in the areas (Carpenter and Miller l98la), but it is not clear why the CIC index escalated by an order of magnitude. The BWE index for l979 included significant contributions from the wild- life and endangered species modules. If these are removed, the Q index drops from 86.2 to 39.8, making BWE more environmentally desir- able than OPM but less so than CIC. There is no indication of what range of statistical variations the indices include, nor how signifi- cant the differences among Q values for different trials might be. Endangered and Threatened Species The most significant environmental impact of a beltwide boll weevil management program would be the direct and indirect effects of insecticides on non-target organisms. Aquatic biota are clearly the most sensitive, and natural ecological processes would ultimately move and concentrate insecticide residues in streams, rivers, and estuaries of the Cotton Belt. Endangered species are a special concern; fish, molluscs, reptiles, and amphibians constitute the majority and, with the exception of one snail, all are inhabitants of aquatic ecosystems. The toxicology data summarized in Tables 5.8, 5.9, 5.l0 provide no values for molluscs, reptiles, or amphibians. Saltwater crustacea and freshwater crayfish, although not endangered, would also have to be examined closely prior to any beltwide program because of their high sensitivity to diflubenzuron (about 0.003 ppm). Unresolved Issues The information provided to the NRC Committee by the USDA envi- ronmental evaluation team is almost exclusively data generated during the three-year trials. The NRC Committee believes that a truly comprehensive environmental evaluation would have to rely as well on the substantial body of published scientific data on insecticides. The USDA environmental evaluation team did not adequately review and evaluate the extant data.

96 C X! 41 O 0) Cn > C 0) .H id co ^i x! CD id O X2 C 0) 0) 0) 01 .rH x: x! 3 .P .P W . • id X-« X-. ^— ^-» *—-, M Tl' rH CN CO H xT ch o Cft CN ^* rH ^ CO CN rH VD f» ^ fl iH H H s i CQ 1 U 0) •H u ^ •^ rH CN C*) H C0 •*** ^^ »w^ ^«x 3 'H XI 8 4J VO ^* CO *3* id r- ^ CN f*> <y* 4J iH CO iH CN 1 CO id •0 ^H ^ 0) ,i^ C0 c 1 u <d •H a X-v *— X ^^ **'x 1 rl TJ' rH CN ro 0) 0) U d C X 0*^ r-i f-t t~- rH •H o (•*» O « O CN u rH t rH CN <N H* H z * "2 rH CO o. 2 II) . H |5 X! ^^ 0 ^—^ IB .— '* ^«« ^-~v ^— . 4J ^ H o M ^' CN rH f) f-~, 00 co «-• >-- *«* «*^ **^ g 0^ cn on ^^ o p•l rH rH CO r** o r** o> C (>» in VD CO iH IH o^ <N rH CN ^ M M rH 1 44 <D iH 0) H c c rH rH -H 0) •H •H ^^ id ad 1 ^ ~ ^. 11 X) o u 1 1 M >•^ %^ ^^ •*«- " C0 •d id 0) g M 0> CO LO (N 0^ fl? f^. 1 t-H ^ +5 0^ CO <£> «* 1 id M C C H VO CO rH (0 OJ H & u H rl •H •H <d td ^1 9B u u s i •H 3 • • • •H H 2 Cft J^ u 1 § •H M O CO B id •H X! OHJ •H M CO 'P C0 t-l § £-i •H O CO S W U £ 02 P4 S H H O ffl U U

97 A fundamental shortcoming in the team's report (APHIS l98la) is that the trials were not designed to measure environmental effects and thus obtain the data necessary to predict the environmental effects of beltwide OPM or BWE programs. The primary emphases of the trials were biological effectiveness and economic practicality. This criticism alone does not mean that a beltwide program would have unacceptable environmental risks, but simply that the trials provided inadequate information upon which to base a beltwide environmental assessment. The critical environmental question is whether a belt- wide program would cause significant environmental deterioration. The exposures of human beings, wildlife, and fisheries resources to insecticides cannot be accurately estimated from the information provided by the trials. Models for potential ingestion would have to use data on insecticide degradation, persistence, and toxicity that were specific for each class of insecticides being considered. It is not adequate to consider the ingestion of agricultural products as the only route of exposure—the inhalation of aerosols and suspended soils, the use of surface waters, and the ingestion of dairy prod- ucts, poultry, domestic and wild animals, and fish would also have to be evaluated. USDA reviews of the OPM and BWE trials essentially did not reveal any significant environmental impacts (APHIS l98la); no obvi- ous, short-term, acute effects were observable. But the data are not adequate to demonstrate in unqualified terms a lack of impact or to detect potential long-term and chronic environmental impacts. Because no definitive statements on the degree and kind of environ- mental effects can be made, no realistic estimates of the environmen- tal risks have been incorporated in the calculations of the economic costs of beltwide eradication. The sole measure of environmental risk is that calculated by the BOLL-l simulation model, which provides a relative impact index, Q (Carpenter and Miller l98la). The precision of this index is ques- tionable, since there are no associated measures of uncertainty. The range of Q values is five to tenfold between l979 and l980 for OPM, BWE, and CIC, respectively. These deviations are, in part, due to different insecticide applications in each trial area in different years, but they are also a result of fundamental differences in the contribution of the various indices in different years. The fact that year-to-year variations in Q for the same trial exceeded, in most cases, the differences between the trials makes the usefulness of this approach doubtful. Conclusions Beltwide boll weevil/cotton insect management programs would pose a number of potential environmental hazards which were not evaluated in the USDA risk analysis (Economics and Statistics Service l98la). A number of environmental issues (Table 5.l3) should be addressed region by region before any program is adopted. Much of the information necessary to do so (Table 5.l4} can be obtained from

98 TABLE 5.13 Factors which need to be considered in developing a beltwide environmental assessment of boll weevil eradication from cotton production regions of the United States. Land area involved Acreage Land-use categories Categorization by drainage basins Endangered and threatened species Soil classifications/credibility Regional precipitation Water discharge statistics Quantities of chemicals utilized Cotton management alternative applications Acreages involved, temporal applications Inventory of pesticides to be used Analysis of chemical persistence on various media Environmental fate Suspended solids in rivers Residence time in soils Water/sediment exchange coefficients Food chain concentration factors (aquatic & terrestrial) Erosion potential by soil classification Sedimentation rates in rivers and estuaries Ecological toxicology Toxicity thresholds Toxicology to aquatic resources (also estuarine) Toxicology to terrestrial resources Deleterious effects on natural environment Effects on beneficial insect fauna Regional ecology Biological control alternatives Genetic engineering of cotton varieties Alternative hosts - boll weevil refugia Presence of regional aquatic and terrestrial economic resources Compliance with environmental legislation Clean Air Act Clean Water Act Toxic Substances Control Act

99 Table 5.l4 Example of extant data resources available for a beltwide environmental assessment of boll weevil eradication from cotton production regions of the United States. Cotton production (by Water Resource Regions) Acres planted Acres harvested Lbs/acre yield County statistics • % cropland • % cotton • % other Yield with and without irrigation Yield by county Land use Cotton (lbs/acre per county) Land use - rangeland/pasture Land use - cropland Land use - urban Land resource regions Land capability classes Insecticides in the environment Residues in water (EPA) Residues in soils/sediments (USDA) Residues in food (FDA) Monitoring of bioaccumulation in biota Monitoring of residues in human adipose tissues Climate Annual rainfall Moisture index Evapotranspiration index Annual temperature Min. annual temperature Max. annual temperature

l00 Table 5.l4 (Continued) • Soils Erosion index vs rainfall Erosion index vs % cropland in cotton counties Erosion index vs % county in cotton Types, agricultural suitability classifications • Water resources Average runoff No. of lakes/state Major rivers/estuaries Principal drainage basins Surface water runoff Season's highest streamflow Season's lowest streamflow Water resource regions Suspended sediments Annual turbidity Water quality • Unique resources by Water Resource Region Rare and endangered species Wilderness areas Proposed wilderness areas National forests National parks National wildlife refuges National scenic rivers Research national areas Nature conservancy areas

l0l existing sources. The future trends of a CIC program, an OPM-BWE program with continued maintenance and the establishment of a buffer zone, and a publicly managed 0PM program are the three alternatives that would require environmental evaluation. As CIC improves, the technological differences between OPM and CIC will disappear with time. Initially, however, an OPM program would have environmental advantages over CIC in some regions of the Cotton Belt. Projections of environmental impact should also consider trends in cotton produc- tion in the infested regions if no beltwide plan is implemented. Reduced cotton acreage and improved private cotton insect control practices might reduce the present environmental impacts. The two major types of environmental effects appear to be (l) those related to changes in insecticide use patterns, and (2) those related to changes in cotton acreage. It should be possible to construct a baseline estimate of the total insecticide load in the Cotton Belt region by region. Projecting this estimate to cover the years l980-2000 would be more difficult but would seem to be neces- sary for evaluating the alternative strategies. The estimation process should include insecticide residue concentrations in air, water, sediments, soils, foods, human adipose tissue, and selected biota on a regional basis. These projections should be stochastic to allow for variations in space and time. The environmental effects of implementing a beltwide OPM or BWE program should be evaluated as alternatives to the projected environmental effects of CIC practices. A BWE program (the OPM-NI-BWE option) would be expected to result in increased chemical concentrations initially but reduced concentrations over time except in the Texas buffer zone, where high rates of insecticide use would continue, and in those parts of the southeast where frequent spraying is required to control Heliothis. USDA has not clearly determined whether biota would be adversely affected during the initial eradication phases, but the shellfish industries would be placed in a particularly sensitive position. Whether they would survive the initial phases of the program and benefit from the reduced phase is a question that has not been answered technically, even though advocates of eradication draw this conclusion without evidence (Economics and Statistics Service l98la). Similarly, aquatic systems would need to be evaluated in terms of changes in the sediment load of insecticides. Socioeconomic pressures would obviously be important factors in projecting the size of these loads. Some clear environmental benefits would accrue from generally reduced insecticide use. Beneficial insects and spiders would become more important parts of pest management programs, and aquatic life in streams and rivers (especially in the vicinity of fields heavily sprayed in previous years) should benefit from decreased use. The estuarine systems of the Atlantic and Gulf Coasts might also respond positively to reduced insecticide use. Wildlife in terrestrial systems also should benefit. Two environmental benefits appear likely to accrue to human beings directly. One would be the de- creased exposure of insecticide applicators and other field workers, while the other would be a reduced concentration of insecticides in human foodstuffs and tissues. While these are potential benefits, they have yet to be demonstrated as reasonable probabilities.

l02 SOCIOLOGICAL IMPLICATIONS The probability of successfully carrying out an eradication or OPM program would rest heavily on human behavior. Beltwide implemen- tation of an eradication program would require legislative action at the state level, and the economic and environmental factors that caused North Carolina farmers to accept eradication regulations (by means of a referendum) might not exist in Texas and thus influence the outcome of a referendum there. Without the necessary regulatory authority in each state, eradication would be impossible. On the other hand, areawide pest management would not require total elimination of the weevil to be successful. The OPM trial in Mississippi demonstrated the value of a pest management program with a publicly funded incentive, and this carrot-and-stick approach might be more successful in obtaining nearly l00 percent participation among growers than an educational approach. But in either case, grower acceptance would be much more likely if the program selected for implementation allowed flexibility and individual choice, which are always more acceptable to the American farmer than governmental decrees. The data developed in the OPM and BWE trials clearly show that area programs are more successful in maintaining low boll weevil populations than the "each-grower-do-his-own-thing" approach (CIC). Thus, it would seem logical to consider various scenarios that might provide the impetus for grower acceptance of a beltwide OPM or BWE program. Optimum Pest Management The approach used in Panola County was a good example of farmer- government cooperation to achieve a goal. The use of a subsidy, maintained for 3 years in the form of compensation for weevil dia- pause treatments, was extremely successful. It suggests that such a model would be useful elsewhere in the Cotton Belt to obtain areawide participation. The cessation of the subsidy will mean that a certain percentage of growers will cease making diapause treatments. How large a per- centage will do so remains to be seen, but the benefits of the trial should persuade most of the growers, perhaps 75 percent, to continue doing so. If weevil populations begin to resurge, growers who ceased making treatments would logically resume making them for economic reasons. Pest Management Associations and Cooperatives Pest management associations and cooperatives, and private pest-management consultants, have successfully organized and handled cooperative cotton insect management programs in Texas, where cotton- growing areas have more environmental diversity than those of any

l03 other state in the Cotton Belt. Texas is the largest cotton- producing state in the boll weevil infested area of the Cotton Belt, but there are many areas in the state where the weevil is not a major pest. These areas are the least likely to receive great benefits from weevil eradication. Thus, Texas may be the most difficult state in which to obtain grower approval of a mandatory weevil eradication. The state Extension Service had a great deal to do with organi- zation of the Texas Pest Management Association, which, along with private consultants, has been very active in bringing about grower acceptance of crop management practices that will provide the best control of cotton pests. All of these practices are economically sound and lead toward reduced insecticide use. The extension service in other states has also fostered areawide approaches to cotton insect management. Such approaches can be carried out in any area of the Cotton Belt, limited only by the desire and imagination of the local populace. Current Insect Control (CIC) Current insect control (CIC) is not without its success stories. Pontotoc County, the comparison area for the OPM trial, has been used as an example of successful CIC, and in some other Mississippi coun- ties, 90 percent of the cotton acreage is under a pest management consultant's care (J. Kimbrough III, Lexington, MS, personal communi- cation, l980). In areas where private consultants are particularly active, growers utilizing their services have seen the economic and ecologic advantages of coordinated pest management in practice. The adoption of more sophisticated pest management practices because of greater environmental and economic awareness among growers is in- creasing each year. Current insect control practice is rapidly evolving into integrated pest management (IPM), which lacks only government subsidies for diapause treatments to become optimum pest management (OPM).