Demographic responses of Daphnia magna fed transgenic Bt-maize

Daphnia magna fed Bt-transgenic maize showed a reduced fitness performance as compared to animals fed the near isogenic unmodified control maize grown in the same environment. The addition of predator smell increased the differences between the feeding groups, indicating an interaction between food and predator treatment.

The H₀-hypothesis (i.e. Cry1Ab-maize and near isogenic line are substantially equivalent) was rejected. On a biological level, as food sources for D. magna, the two maize varieties were not performing similarly: (1) the survival of the GM-fed animals were significantly lower than UM-fed, without predator (p = 0.037, coxph-test), but not significantly different for the group with predator or for the total material (p = 0.45 and p = 0.16, respectively, coxph-test); (2) the total number of eggs produced by GM-fed animals were lower as compared to animals fed on UM-maize; (3) the growth showed a trend towards larger body sizes in the UM-fed group without predator smell present (Bøhn et al. 2008). When the predator smell was included the UM-fed animals had a significantly larger body size at day 29 and 42 in the experiments; (4) the population growth rate, combining survival and fecundity data, showed a higher growth rate in the UM-fed animals with non-overlapping 95% confidence intervals for both the predator groups (n = 20) and for the total material (n = 80). These results sum up differences in repeated and fully randomized life-cycle experiments, using maize as the only source of food.

The most common definition of substantial equivalence has emphasized the biochemical components of the plant (Brake and Vlachos 1998; Clark and Ipharraguerre 2001; Sidhu et al. 2000), thus potentially overlooking important elements like post-translational modifications and immune effects (Prescott et al. 2005), insertional effects leading to up- or down regulation of endogeneous genes (Filipecki and Malepszy 2006), abberations in transgene-specific mRNAs (Rosati et al. 2008), or subtle changes in the proteome and metabolome (Manetti et al. 2006; Zolla et al. 2008). Animal feeding studies are performed to reveal subtler effects on survival, growth, histological indicators, and other characteristics of consumers. This means that effects can be picked up even without a clear understanding of the cause(s), e.g. the mode of action for a toxin.

Our data demonstrate negative effects in the non-target organism D. magna feeding insecticide maize, but we lack the mechanistic explanation for the effect. A number of other studies, mainly in terrestrial invertebrates, have also showed negative effects of Bt-transgenic plants on non-target species. Lövei and Arpaia (2005) found significantly negative effects on about 30% of the studied invertebrate predators and 57% of the studied parasitoids after feeding Bt-plants or pollen. Hilbeck and Schmidt (2006) report 27 out of 54 studies (50%) that showed negative effects after feeding Bt-plants or pollen. Ramirez-Romero and co-workers showed a tri-trophic effect where a parasitoid hymenopteran species was negatively affected (developmental time, adult size and fecundity) via host exposure to Cry1Ab proteins and Bt-plants (Ramirez-Romero et al. 2007). The same group also showed that honey bees were negatively affected in their feeding behaviour and that their learning performance was reduced when exposed to 5,000 ppb of Cry1Ab protein (Ramirez-Romero et al. 2008). Moser and co-workers showed that the developmental time of C. maculata increased after Bt hybrid corn treatments compared with non-Bt corn treatments (Moser et al. 2008). In the aquatic environment, it has been demonstrated that toxin-containing crop byproducts from maize are dispersed, decomposed and consumed by stream insects (Rosi-Marshall et al. 2007). When fed Bt-maize in the laboratory, two non-target caddisflies showed higher mortality and reduced growth (Rosi-Marshall et al. 2007).

None of the mentioned studies were able to identify the cause(s) of the effect(s), e.g. the mode of action for the Bt-toxins. This leaves us with a knowledge gap and an urgent need for more detailed studies focusing on mechanism. New methodological techniques like functional genomics, transcriptomics, proteomics, metabolomics, etc. may prove helpful in order to improve the understanding about both the properties of the plant, and the properties of the test organism that feed on that plant. This will immediately bring the discussion well beyond the coarse assumptions made by the substantial equivalence principle.

One promising research direction is ecotoxicogenomics where gene expression patterns are studied under the impact of different treatments, e.g. exposure to chemicals. Watanabe and co-workers showed that chemicals expected to impact oxidative stress, respiratory uncoupling, metabolic change and heavy metal exposure altered gene expression patterns in D. magna in recognizable ways (Watanabe 2007). Also, DNA microarrays for D. magna have been used to link gene responses to the impact of toxicants on life-history traits and population growth rate in D. magna (Connon et al. 2008). If, or when further studies are able to link a negatively affected phenotype after exposure to e.g. Bt-toxin, or after feeding on a transgenic plant, to up- or down-regulation of certain genes or pathways, a new and promising step towards finding modes of action would have been established. However, a functional genomics determinism, i.e. tightly relating changes in microarrays, changes in proteins and changes in the metabolome is far from established. Assuming such determinism could narrow our investigations and understanding of the phenotype.

In spite of the abovementioned, the basis for analysis of risks from transgenic plants still largely remains that of substantial equivalence. Following Millstone and co-workers that called the Substantial Equivalence Principle a regulatory principle without any biological relationship or theoretical validity (Millstone et al. 1999), we argue that the concept of substantial equivalence should have a definition based on its biological activity. It is the biological activity that is relevant in the agro-ecosystem, in relation to non-target effects and in a human health context. Otherwise, there would be no necessary link between substantial equivalence and safety.

Hypothesis supported. Given the differences in the performance of D. magna fed GM versus UM maize and the detailed demographic analysis, we were able to evaluate which stages of the D. magna life cycle were the most sensitive to differences in food quality, and to predator smell. The survival curves of both feeding groups were relatively straight and the fecundity tended to increase with age. However, the sensitivity analysis revealed that what really mattered to the population growth rate came from differences in survival in juveniles and young adults and from fecundity differences in young adults. Thus, the structured population models offer results on sensitivity which may have important implications: juveniles and young adult stages are the most sensitive experimental units and should be prioritized in toxicological/ecotoxicological research (Connon et al. 2008; Forbes and Calow 1999). This is often contrary to established practice, where well fed adults are used for testing. For example, Cry1Ab pollen is tested in adult D. magna, short term with no treatment related effects found (Mendelson et al. 2003). In short-lived test organisms, we recommend the use of life cycle experiments to enable stage specific evaluation of toxicity and general life-history performance, including population level comparisons. When applicable, these methods are superior to traditional testing practices.

Hypothesis partly supported. A trade-off response was observed in animals exposed to predator, but only in the UM-fed animals. This group increased their population growth rate by increasing their early reproduction as anticipated by life-history theory (Hansen et al. 1999; Mauri et al. 2003; Twombly et al. 1998). However, the early boost in reproduction coincided with a relatively high mortality, indicating a cost of the response. When a predator is present, the expectation of a shorter life span may trigger early reproduction (Sakwinska and Dawidowicz 2005). Also other stress factors than predation may lead to high fecundity early in life (Coors et al. 2004) at the cost of future quantity or quality of offspring, e.g. due to a compromised immune system (Hanssen et al. 2005; Sheldon and Verhulst 1996). In general, a negative correlation between current and residual reproductive value, or between other life-histoty traits, has been used to show the cost of reproduction (Harshman and Zera 2007; Stearns 1992).

Interestingly, the GM-fed animals showed no apparent response in fecundity to the presence of the predator cue. We argue that the lack of the anticipated response and the relatively low reproductive output for this group was caused by a reduced health condition after feeding the Bt-transgenic maize. The appropriate life-history response may simple require a reasonably good health condition. The lack of response by early reproduction may, however, explain the somewhat higher survival in the animals fed Bt-transgenic maize and exposed to predator, compared to the early reproducing animals that were fed unmodified maize and exposed to predator.

Looking at the population level, the predator smell stimulated a higher population growth rate for the group fed UM-maize, which may seem paradoxical since a predator usually represents a stress factor that reduces growth and delays reproduction (Auld and Relyea 2008; Brodin et al. 2006). However, as the experiments did not last long enough to detect possible counteracting responses towards the end of life, i.e. a reduced reproductive output after the end of the experiment (day 42), the expected trade-off between early and late reproduction could not be evaluated completely. Also, the predator smell may have contributed with a small portion of nutrients, e.g. bacteria, to the sterile ADAM medium, thereby stimulating the test animals.

Hypothesis supported. The differences between the GM-fed and UM-fed animals were larger in the presence of a predator: the population growth rate was markedly reduced in animals exposed to both GM maize and predator as compared to animals exposed to UM maize and predator. Also, the differences in body size for GM and UM-fed animals were much clearer in the groups with predator treatment, c.f. (Bøhn et al. 2008). Thus, the results indicate that the effects of food treatment and predator treatment were not additive, but interacted. As argued above, we emphasize that an organism under stress from two factors may be unable to respond adequately to several stressing conditions in the environment.

In conclusion, this study shows negative effects of a Bt-transgenic maize variety on a non-target model organism. The study thus rejects the substantial equivalence of the tested GM and UM-maize. Extended analysis using structured population models combine effects on survival and reproduction and enabled us to compare effects both on the population level, and on the sensitivity of different age-groups within food and predator treatments. Life-cycle experiments analysed with structured population models should be used, when applicable, to derive better estimates of the impact of novel food stuff, pesticides, toxicants and other stressors. Such demographic studies offer complete time-series portraits of toxicity and represents an improvement over traditional toxicological data and analysis. We urge more research on altered phenotypes and life-histories due to exposure from transgenic plants, ideally in combination with detailed—omics techniques, to follow the missing link of the specific causes and mechanisms involved in observed negative effects.