Interpreting the Scientific Literature. Differences in the Scientific and Lay Communities

May R. Berenbaum Member of the National Academy of Sciences That the lay public can react differently to the scientific literature than does the scientific community was dramatically demonstrated recently in the context of genetically manipulated organisms. The journal Nature, a high-profile journal with a reputation in both the lay and scientific communities for quality, includes, along with full-length reports, a section called “Scientific Correspondence,” articles that are preliminary in nature. On May 20, 1999, a scientific correspondence appeared by John E. Losey, Linda S. Rayor, and Maureen E. Carter, titled “Transgenic pollen harms monarch larvae.” This short paper, less than a page in length and accompanied by only a single two-part figure, reported the results of a laboratory study in which pollen from corn (Zea mays) plants containing genetic material from Bacillus thuringiensis (Bt) was applied to leaves of milkweed (Asclepias curassavica) and administered to 25 larvae of the monarch butterfly (Danaus plexippus). Survival, weight gain, and proportion of leaf area consumed by larvae consuming untreated milkweed leaves were observed after 4 d and compared with that of larvae consuming milkweed leaves treated with corn pollen from an “unrelated, untransformed hybrid.” These authors reported that survival, consumption rate, and weight gain were lowest on foliage treated with Bt pollen. These authors then suggested that their results had “potentially profound implications for the conservation of monarch butterflies” due to the fact that milkweed plants, exclusive hosts for monarch larvae, can frequently be found in the midwestern United States around the perimeters of corn fields and thus, depending upon timing, “may be within range of corn pollen deposition” (Losey et al., 1999). To some extent, the findings by Losey et al. (1999) were not particularly novel; the Bt endotoxin gene with which Bt corn had been transformed was known to act specifically against lepidopterous larvae. The target species for these corn events is the European corn borer (Ostrinia nubilalis; Lepidoptera: Pyralidae). Previous reports in the literature of the toxic effects of CryIA(C) protein on nontarget Lepidoptera preceded the Losey et al. (1999)report; for example, Sims (1995) reported on the effect ofB. thuringiensis var kurstaki [CryIa(C)] protein expressed in transgenic cotton on the tobacco hornworm (Manduca sexta). According to Charles Benbrook, a consultant to environmental groups and former associate of the Board on Agriculture of the National Academy of Sciences, “The report had nothing unexpected to those who have studied Bt and Lepidoptera” (Moffat, 2000a). However, it could be argued that the study demonstrated a need to consider more seriously the contributions made by corn pollen in distributing endotoxin beyond a corn plant. Reaction to the Losey et al. (1999) report in the scientific literature was swift and on balance critical. In the June 3, 1999, issue ofNature, John E. Beringer, chairman of the United Kingdom Advisory Committee on Releases to the Environment, published a letter in the “Correspondence” section criticizing the Losey et al. (1999)article on methodological grounds and restating its “preliminary rather than definitive” nature. Beringer (1999) specifically pointed out that the experiment lacked a critical control treatment—i.e. milkweed leaves treated with pollen from the same variety that had been transformed to generate the Bt corn. Although not specifically stated by Beringer, this omission is of importance at least in part because, according to Losey et al. (1999), the presence of untransformed, unrelated pollen on foliage reduced consumption rates significantly.Beringer (1999) also pointed out that the authors had not done a dilution study to determine the level of corn pollen contamination that would, under field conditions, affect larval growth and survival. With respect to the study, this author stated “Of course it is desirable to point out the potential harm that may arise from pollen dispersal… but the data reported by Losey et al. do not directly pertain to this issue… . Preliminary observations should not be overinterpreted. Regrettably, most reporting of the communication has almost entirely ignored the need for such caution.” Another response to the Losey et al. (1999) report appeared in the September 1999 issue of Nature Biotechnology and was authored by A.M. Shelton and R.T. Roush. Its tone was suggested by its title: “False reports and the ears of men.” Shelton and Roush (1999) cited a previous (but unpublished) field study examining Bt corn pollen deposition on milkweed plants in and around corn fields that failed to document significant impacts on monarch larvae. About theLosey et al. (1999) paper, they wrote: “We believe that few entomologists or weed scientists familiar with butterflies or corn production (and the control of milkweed) give credence to theNature article, but the public and its policy makers have reacted in a knee-jerk fashion… . Was this reaction justified based on what can only be considered a preliminary laboratory study or could rumor still be more entertaining than fact?… Are studies such as these guilty of ‘stuffing the ears of men with false reports’ or is it the willingness of people to accept uncritically any reports that fit their own perceptions that is really to blame?” Thus, critics of Losey et al. (1999) agreed that the preliminary nature of the limited laboratory study conducted by Losey et al. constrained its applicability to real-world, field situations. Over and above these limitations, a lack of detailed methodological description in the paper further limited its interpretability. The failure of the authors to identify the precise “unrelated, untransformed hybrid” meant that subsequent investigators could not, without personal contact with the authors, repeat the study exactly. Also preventing precise replication was the failure of the authors to report the exact pollen dosage used. They instead reported that “pollen density was set to visually match densities on milkweed leaves collected from corn fields” without specifying where these leaves were collected or how the visual match was checked for consistency. Criticism of methodology is not unusual in the scientific literature; the conduct of science dictates critical interpretation of previously published studies as well as verification by repetition. Reaction by the press and within the lay community to this paper was, in contrast, overwhelmingly uncritical in that few if any accounts indicated that the Losey et al. (1999) study was inconclusive or preliminary in any way. In an editor's letter that appeared in the popular journalTaste for Life, R. Frost (1999) wrote “Witness recent research that corn genetically altered to contain Bacillus thuringiensis can kill monarch caterpillars… . Genetically spliced into corn, this bacteria [sic] produces a pesticide-like protein, which is contained in the pollen. The pollen has been found to kill substantial numbers of monarch butterfly larvae, according to Cornell University researchers. The fact that killing butterflies is an unintended consequence makes thoughtful consumers wonder what other fallout will result from this and other genetically engineered organisms.” In another example of public reaction, the fall 1999 mail solicitation of the Environmental Defense Fund (EDF) came to contributors in an envelope bedecked with monarch butterflies and emblazoned with the text “First the butterflies… . Then?” Inside, the letter from Executive Director Fred Krupp contained the text, “disturbing new scientific evidence about biologically engineered crops suggests the milkweed the Monarch caterpillars eat may now be threatening their very survival… . Just as Midwestern farmers were sowing eight million more acres of American farmland with new, genetically manipulated corn, a report in the journal Nature threw up an alarm. It showed that these farming operations could be wreaking environmental havoc with each kernel they plant. The report, by three scientists at Cornell University, was a wake-up call to America. And because of its grave implications, I'm writing you today to enlist yoursupport… Like you, we want the food Americans eat to be as safe as possible, and agriculture to have minimal environmental impact on our planet's fragile ecosystems including delicate species like the Monarch butterfly. With your continued support we'll bring common sense to bear on this potentially serious environmental problem before the Monarch butterfly ends up on the Endangered Species list.” It is curious that, although no criticism of methodologies in the Losey et al. (1999) study appeared in the materials distributed by the EDF, such criticisms were leveled at subsequent industry studies. In the December 1999 issue of EDF Letter, a “bimonthly report to members of the Environmental Defense Fund,” an account appeared of a November 1999 conference convened to discuss potential impacts of Bt corn and funded largely by the biotechnology industry (Watson, 1999). EDF scientist Rebecca Goldburg was quoted as saying “Most of the reports were preliminary, sample sizes were sometimes small, and some research methodologies were questionable.” These are basically the same charges that were leveled earlier in the scientific literature at the original Losey et al. (1999) study. At the very least, it can be stated that different standards were applied by scientific readers and lay readers in interpreting the significance of a single study, identified by the journal in which it was published as preliminary. Public reaction was swift and massive; that opposition to Bt corn stemmed largely from the Losey et al. (1999)study was difficult to deny given that protesters appeared at demonstrations staged during public hearings sponsored by the “Federal Drug Administration” dressed as monarch butterflies (Moffat, 2000b). Paul Raeburn, technology editor fromBusiness Week, explicitly “attributed the new hostility [toward bioengineered foods] largely to a laboratory study showing that pollen from GM corn kills the larvae of monarch butterflies” (Steinberg, 2000). The Environmental Protection Agency (EPA) itself acknowledged the influence of the Losey et al. (1999) study in issuing new rules governing the planting of Bt corn, specifying that all fields containing corn engineered with Bt endotoxin have to be provided with non-Bt corn as buffer, with no more than 50% to 75% of any field planted to Bt corn. The EPA press advisory specifically mentioned the need to “protect non-target insects, particularly the monarch butterfly” (Moffat, 2000b), although the principal motivation behind the planting restrictions was to delay resistance acquisition in target species. Also alluding to the Losey et al. (1999) study was the Federal Insecticide Fungicide, Rodenticide Act Scientific Advisory Panel Report Number 99–06, released February 4, 2000, considering “Characterization and non-target organism data requirements for protein plant-pesticides.” Specifically included in the November 9, 1999 charge was the question: “Should OPP [Office of Pesticide Programs at EPA] extend non-target insect effects testing requirements to include secondary exposure scenarios, like pollen covered milkweed and lupine or intoxicated aphids?” This charge was regarded as important (p. 17): “In view of recent publications that indicate secondary exposure may pose ecological risk to certain non-target insects, it is imperative that the Agency assume the responsibility to conduct appropriate ecological risk assessment regarding this controversial issue.” The Panel's conclusion was that current nontarget testing requirements were inadequate, in that they were limited in terms of species numbers, and that test species to be added “might include” nontarget relatives of the target pests. This is despite the Panel's conclusion (p. 17) that “most scenarios for secondary exposure are not very compelling because of the low concentration of active ingredient likely to be delivered to non-target insects or the small fraction of the non-target insect population likely to be affected. The half-life of a pesticidal protein also is expected to be short. Therefore, the Panel suggests that food chain effects are not likely to be significant.” Key to the recommendation, however, was the Panel's perception of the obligation of the Agency to address the public controversy. This scientific Panel called for “additional research… on the various possible effects of plant pesticidal proteins on non-target insects.” Such a call is not at all unexpected in the scientific arena and implicitly suggests that the current scientific appraisal of nontarget risks of Bt corn is that the results are inconclusive and thus should not be used exclusively to determine policy. Much the same perspective was voiced by Losey et al. (1999) in the rarely quoted conclusion of their paper: “it is imperative that we gather the data necessary to evaluate the risks associated with this new agrotechnology and to compare these risks with those posed by pesticides and other pest-control tactics.” Whether the data once obtained will be fairly evaluated is another issue altogether. When a large field of transgenic corn appeared next to a field site where I have worked with my colleagues for almost 20 years studying native populations of black swallowtail butterflies (Papilio polyxenes), our interest in nontarget effects of transgenic pollen became intensely personal. Accordingly, we conducted a field experiment to assess the impact of transgenic corn pollen under field conditions (Wraight et al., 2000). This study was funded entirely by university funds and was motivated strictly by our interest in the study organism, the black swallowtail butterfly. The farmer who had planted the cornfield entirely for his own purposes determined for us the choice of event—he had selected and planted the event presumably because of its agronomic properties and I doubt that he was even aware of the fact that his cornfield provided us with a natural source of transgenic pollen. In our field study, we documented high levels of caterpillar mortality, none of which could be directly attributed to corn pollen; caterpillars set out on pots containing their host plants at considerable distances from the corn pollen source died at rates indistinguishable from those of caterpillars on host plants in pots immediately adjacent to corn pollen sources. A laboratory bioassay confirmed the lack of toxicity of pollen from Monsanto 810 (the event planted in the field), even at concentrations exceeding those that occur in nature, to black swallowtails; however, we did confirm in the laboratory that some events (notably Novartis 176) produce pollen that is acutely toxic to the caterpillars. We initially submitted the manuscript reporting these results to Nature, largely because we felt it was an appropriate venue in view of its publication a year earlier of the Losey et al. study. It was rejected on the basis of a single review, the gist of which is that our report was inconclusive because it was based on a single species (as was the Losey et al. study). The manuscript received a more favorable reception at the Proceedings of the National Academy of Sciences, where it was published, initially online, the year following the study (Wraight et al., 2000). Reaction to publication was swift and depressingly predictable from two sides. On one side, the Wall Street Journal (June 6, 2000) reported that “Study on Corn Should Bolster Biotech Firms,” a conclusion that was hardly merited by our study. On the other side, a scathing critique appeared on the Institute of Science in Society Web site authored by Mae-Wan Ho and titled “Swallowing the tale of the swallowtails: no ‘absence of toxicity’ of BT pollen.” Ho (2000) criticized certain elements of the design and interpretation of the study, at the same time citing as more authoritative a study that existed in published form at the time only as a single-paragraph abstract online. Among the other statements made about our study, Ho (2000) went so far as to call it an “abuse of science.” It is interesting that, although it may still be early (in view of the fact that most journals require more than 6 months between submission and review), Wraight et al. (2000) has failed to elicit the same sort of negative reaction within the refereed scientific literature thatLosey et al. (1999) received. At the moment, it appears that the American scientific community doesn't perceive itself as having been abused. The disparate responses to our study, though, point out the hazards of conducting research in which the general public has an interest. Policy decisions should be made on solid scientific evidence, yet it is usually the lay public that is the source of concern about policies. At the moment, there is likely not enough solid evidence, by scientific standards, to make a rational decision on Bt corn at this point—but there will never be if studies that run counter to a particular agenda are rejected out of hand on criteria that are not applied equably to all studies. Nor will there be if the amount of evidence demanded is for a level of certainty that can never be achieved or for a level of knowledge that can never be obtained. The monarch butterfly is a beloved icon, familiar to millions; it appears in children's books and greeting cards, on T-shirts and jewelry, and in fact has been designated a is the insect of It has been the of study by of yet if this BT corn has pointed out it is the of knowledge that about natural The monarch is but one of at least butterfly species in despite its as an icon, of its a That the known butterfly is an that the of knowledge of the other is even more In a of studies in conducted in demonstrated that on is or 1999). The to this between public and scientific interpretation and of not to conduct research of monarch is That is itself the although not the is to more time and to science in this The decisions made by the public as consumers and in a lack of with the science Academy of Sciences, 2000). A recent National Academy of Sciences study on the of in Agriculture in applied and risk for the moment, such is not only at the level but for at the this same report that be to and scientists about the of public can no with only with it is in the interest of that the public can directly from scientists about their research rather than on or popular with of their and as consuming as it may be to to for or even to for popular it is important for scientists to to this with the same level of and at the conduct of the scientific study policy decisions on how the public scientific and the responsibility of scientists in research findings does not with only with their I my for and thoughtful