Online Supporting Information

ONLINE SUPPORTING INFORMATION

Appendix S3-Background material considered to aid the choice of parameters governing the range of sensitivities of species

A major difference between the maize event 1507 expressing Cry1F protein and the maize event MON810 expressing Cry1Ab protein is in their respective Bt protein content in pollen, which directly affects toxicity. The content of the Cry1F protein in maize 1507 pollen has been estimated as 32 ng/mg dry weight diet (US EPA, 2001, 2005) and for maize MON810 pollen as about 0.09 ng/mg dry weight (Mendelsohn et al., 2003). There is therefore more than approximately 350 times the Bt protein content in maize 1507 pollen than in maize MON810 pollen.

The level of expression of Cry1F protein in maize pollen is variable and various values have been reported in the literature (EFSA, 2005; and see also EFSA, 2006 and EFSA, 2008), dependent on the variety used, environmental conditions and other factors. As was the case in Perry et al. (2010), where there was a choice in model structure we have adopted the worse-case option, resulting in greater mortality estimates. This guards against the underestimation of mortality, important in environmental risk assessment. Hence the highest level of expression reported was used here. As an indication of this difference in toxin expression, for maize MON810 Perry et al. (2010) assumed that the LC50 value for fourth instar larvae of the moth Plutella xylostella was 3626 maize pollen grains cm-2, whereas for maize 1507 the LC50 value reported by Wolt et al. (2005) for first instar larvae of Plutella xylostella was 54 maize pollen grains cm-2. Indeed, most of the LC50 values considered here for Cry1F are considerably less than those considered by Perry et al. (2010) for Cry1Ab (see below).

Wolt et al. (2005) reported LC50 values for Cry1F for 16 lepidoptera, mainly pest species, from an unpublished technical report of Dow AgroSciences LLC, showing a wide range of sensitivities from 0.065 – 410 μg Cry1F g–1 diet, with a geometric mean LC50 of 10.3 μg Cry1F g–1 diet (which they calculated as approximately equivalent to c. 8500 maize pollen grains cm-2 leaf). Wolt & Conlan (2001), in an unpublished technical report of Dow AgroSciences LLC, fitted a normal distribution to those data and then extrapolated the lower value to give an estimate of the 10th%ile of the sensitivity distribution as 553 grains cm–2, an effect level which they regarded as conservative representation of the effect endpoint for the tier 1 risk assessment for a hypothetical sensitive species of concern.

Wolt et al. (2005) then extrapolated the lower value to give an estimate of the 5th%ile of the sensitivity distribution as 0.04 μg Cry1F g–1 diet (33 maize pollen grains cm–2 leaf), an effect level which they regarded as representing a worst-case effect endpoint for the tier 1 risk assessment for a hypothetical sensitive species of concern. However, as Wolt et al. (2005) stated explicitly concerning the uncertainty of their estimates: “these data are representative values based only on exploratory assays, and do not necessarily represent definitive values”.

We gathered estimates of the LC50 for Cry1F and for Cry1Ab for 20 different lepidopteran species from various sources: Gaspers et al. (2010); González-Cabrera et al. (2006); Hellmich et al. (2001); Monnerat et al. (1999), Sayyed & Wright (2001); Siqueira et al. (2004); Wolt et al. (2003); Xu et al. (2010), in addition to Wolt et al. (2005). These estimates are subject to greater variability than those of Wolt et al. (2005) because: (1) the studies from which they are derived have taken place in different environments; (2) the studies used different populations of larvae, even when the same species is involved; (3) involved different batches of toxin; (4) used activated or protoxins; and (5) adopted different methodologies, such as surface application and diet incorporation. Monnerat et al. (1999), Saeglitz et al. (2006), Schuphan (2006) and Gaspers et al. (2010) are amongst an increasing number of authors who have warned about the additional variability induced by these and other factors. Nevertheless, the data suggest that the calculated geometric mean LC50 of 10.3 μg Cry1F g–1 diet (c. 8500 maize pollen grains cm2 leaf) reported by Wolt et al. (2005) might be a slight overestimate, although we consider it would be inadvisable to attempt a more accurate estimate. Also, it is clear that the relative sensitivity of species to Cry1Ab compared to Cry1F varies considerably. For example, larvae of Danaus plexippus, the Monarch butterfly, are known to be relatively insensitive to Cry1F; similarly for Sesamia nonagroides, the Mediterranean corn borer. However, for larvae of Spodoptera frugiperda, the Fall Armyworm moth, the reverse appears to be the case. Whilst, on average, Lepidoptera are five times less sensitive to Cry1F than to Cry1Ab, there is considerable variability between species.

It is to be expected that variability in LC50 values for Cry1F will be inflated by factors including those discussed above. We note the methodology adopted by Wolt et al. (2005) utilized the 5%ile, and that this choice is somewhat arbitrary and in a more conservative approach this might be replaced by a 1%ile or some other smaller value. In this regard it is relevant to consider the number of non-target Lepidoptera that might be potentially exposed within a maize arable ecosystem. In Austria (UmweltBundesamt, 2007) considers that over 150 butterflies may be potentially exposed. The Rothamsted Insect Survey regularly identifies over 600 macro-lepidopteran moths in light-traps across the UK, although many of these traps are in arable habitats. In a list of 500 species, we would expect about 24 species to be more sensitive than the lower 5th centile. By contrast, the most sensitive species would be expected to have an LC50 that was of the same order of magnitude as the lower 0.2%ile of the species sensitivity distribution.

We have therefore adopted a range of sensitivities to Cry1F in our model that includes, at the lower end, a more pessimistic worst-case effect level than the value of 33 (maize pollen grains cm2 leaf) assumed by Wolt et al. (2005). These values are intended to represent a range of different, but unspecified lepidopteran species and thus to reflect the between-species variability in acute sensitivity to Cry1F. The five LC50 values (maize pollen grains cm2 leaf) we consider are a geometric series with 11.4-fold increments: 1.265, 14.36, 163.2, 1853 and 21057. The smallest value for the five hypothetical species, 1.265, may be considered as representing the ‘worst-case’, where ‘extreme’ sensitivity to Cry1F protein from maize 1507 would bring the greatest risk of mortality to a non-target lepidopteran species. This corresponds very closely to the estimated lower 0.2%ile of the species sensitivity distribution. The next smallest value, 14.36, represents a very-highly sensitive species corresponding fairly closely to the 1st percentile. The highly-sensitive value, 163.2, lies between the 5th and 10th centile of the distribution. The value, 1853, is highly likely to be below the mean and is termed ‘below-average’. The value, 21057, is highly likely to be above the mean of the distribution and is termed ‘above-average’.

The ranked LC50 values reported by Wolt et al. (2005) are shown in units of maize 1507 pollen grains cm-2, together with the values studied here, in the Table below, both on the natural and the logarithmic scale.

Table Sensitivity of first instars of various lepidopteran species (expressed as LC50 values in units of maize 1507 pollen grains cm-2) to the Cry1F protein, together with corresponding values studied here

(After Wolt & Conlan (2001) and Wolt et al. (2005). Lower centiles were estimated from a normal distribution fitted to the distribution of species sensitivity).

Species / Categorisation / LC50 / Log10(LC50)
Hypothetical species / Worst-case extremely sensitive / 1.265 / 0.10
0.2%ile (estimated from Wolt & Conlan, 2001) / 1.27 / 0.10
1%ile (1st percentile; estimated from Wolt & Conlan, 2001) / 13.8 / 1.14
Hypothetical species / Very highly sensitive / 14.36 / 1.16
5%ile (5th percentile; Wolt et al., 2005) / 33 / 1.52
Plutella xylostella / 54 / 1.73
Hypothetical species / Highly sensitive / 163.2 / 2.21
Ostrinia nubilalis / 479 / 2.68
10%ile (10th percentile; Wolt & Conlan, 2001) / 553 / 2.74
Spodoptera littoralis / 817 / 2.91
Heliothis virescens / 1551 / 3.19
Trichoplusia ni / 1774 / 3.25
Hypothetical species / Above-average sensitivity / 1853 / 3.27
Spodoptera frugiperda / 1980 / 3.30
Spodoptera exigua / 6435 / 3.81
Crambus spp. / >8250 / >3.9
Mean of distribution as estimated by Wolt et al. (2005) / 8497 / 3.93
Hypothetical species / Below-average sensitivity / 21057 / 4.32
Spodoptera litura / 22275 / 4.35
Danaus plexippus / >24750 / >4.4
Mamestra configurata / >29700 / >4.5
Diatraea grandiosella / >41250 / >4.6
Agrotis ipsilon / 57090 / 4.76
Helicoverpa armigera / >82500 / >4.9
Choristoneura fumiferana / 115500 / 5.07
Lymantria dispar / 338352 / 5.53

References in this Appendix that do not appear in the main paper

UmweltBundesamt (2007) Review of scientific evidence including latest findings concerning Austrian safeguard measures for GM-Maize lines MON810 and T25. http://www.bmg.gv.at/cms/home/attachments/9/0/3/CH1052/CMS1161157975708/internetversion_1_07_importverbot.pdf

EFSA [European Food Safety Authority] (2005) Opinion of the Scientific Panel on Genetically Modified Organisms on a request from the Commission related to the notification (Reference C/ES/01/01) for the placing on the market of insect-tolerant genetically modified maize 1507 for import, feed and industrial processing and cultivation, under Part C of Directive 2001/18/EC from Pioneer Hi-Bred International/Mycogen Seeds. The EFSA Journal, 181, 1-33. http://www.efsa.europa.eu/en/scdocs/doc/181.pdf

EFSA [European Food Safety Authority] (2006) Clarifications of the Scientific Panel on Genetically Modified Organisms following a request from the Commission related to the opinions on insect resistant genetically modified Bt11 (Reference C/F/96/05.10) and 1507 (Reference C/ES/01/01) maize. http://www.efsa.europa.eu/en/scdocs/doc/181ax1.pdf

EFSA [European Food Safety Authority] (2008) Request from the European Commission to review scientific studies related to the impact on the environment of the cultivation of maize Bt11 and 1507. The EFSA Journal, 851, 1-27.

González-Cabrera, J., Farinós, G.P., Caccia, S., Díaz-Mendoza, M., Castañera, P., Leonardi, M.G., Giordana, B. & Ferré, J. 2006 Toxicity and mode of action of Bacillus thuringiensis Cry proteins in the Mediterranean corn borer, Sesamia nonagroides (Lefebvre). Appl. Environ. Microbiol. 72, 2594-2600

Monnerat, R.G., Masson, L., Brousseau, R., Pusztai-Carey, M., Bordat, D. & Frutos, R. (1999) Differential activity and activation of Bacillus thuringiensis insecticidal proteins in Diamondback moth, Plutella xylostella. Curr. Microbiol., 39, 159-162.

Sayyed, A.H. & Wright, D.J. (2001) Fitness costs and stability of resistance to Bacillus thuringiensis in a field population of the diamondback moth Plutella xylostella L. Ecol. Entomol., 26, 502-508.

Schuphan, I. (2006) Final Report of ProBenBt, Workpackage 1, Task 4. See http://www.bio5.rwth-aachen.de/german/downloads/EU-Review.pdf

Siqueira, H.A.A., Moellenbeck, D., Spencer, T. & Siegfried, B.D. (2004) Cross-resistance of Cry1Ab-selected Ostrinia nubilalis (Lepidoptera: Crambidae) to Bacillus thuringiensis δ-endotoxins. J. Econ. Entomol., 97, 1049-1057.

Xu, L., Wang, Z., Zhang, J., He, K., Ferry, N. & Gatehouse, A.M.R. (2010) Cross-resistance of Cry1Ab-selected Asian corn borer to other Cry toxins. J. Appl. Entomol., 134, 429-438.

1