Season-long expression of Cry1Ac and Cry2Ab proteins in Bollgard II cotton in Australia

Crop Protection 44 (2013) 50e58

Contents lists available at SciVerse ScienceDirect

Crop Protection journal homepage: www.elsevier.com/locate/cropro

Season-long expression of Cry1Ac and Cry2Ab proteins in Bollgard II cotton in Australia Kristen Knight a, *, Graham Head b, John Rogers c a

Monsanto Australia, PO Box 92, Harlaxton, Queensland 4350, Australia Monsanto Company, 800 North Lindbergh Blvd., St Louis, MO 63167, USA c Research Connections and Consulting, PO Box 350, Toowong, Queensland 4066, Australia b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 12 May 2012 Received in revised form 11 October 2012 Accepted 14 October 2012

Bollgard II cotton has been grown commercially in Australia since 2003 for control of the primary target species Helicoverpa armigera (Hübner) and Helicoverpa punctigera (Wallengren) Larvae of both species have been reported to survive at low frequencies on Bollgard II with larvae >8 mm recorded in between 7 and 18% of the area planted to Bollgard II cotton between 2005/06 and 2007/08. F1 and F2 tests have shown that this is not due to the presence of resistance genes in the surviving larvae. To understand if fluctuations in the expression of the Cry proteins in Bollgard II allow some larvae to survive, plant tissue samples were taken from five Bollgard II cultivars throughout the growing season within fields and from different farms within a production region between 2007 and 2010. The data indicate that the expression of both Cry proteins is similar to the known resistance-monitoring diagnostic concentrations and relatively uniform between fields within a farm and between farms within a region, with less than one-third of the tests at this level of variation being significant. However, there were intra-seasonal changes in expression of both Cry proteins and differences in expression between plant structures and between cultivars for both Cry proteins. Further work is needed to establish if this variation in Cry protein content in Bollgard II cotton affects the control of Helicoverpa spp. in the field or whether plant-physiological and pest-behavioural factors underlie the occasional occurrence of Helicoverpa larval survival on Bollgard II cotton. Ó 2012 Elsevier Ltd. All rights reserved.

Keywords: Bacillus thuringiensis Bollgard II cotton Helicoverpa armigera Helicoverpa punctigera Season long control Australia

1. Introduction Bollgard II cotton, Gossypium hirsutum, cultivars expressing the Bacillus thuringiensis variety kurstaki (Bt) Cry1Ac and Cry2Ab dendotoxins have been grown commercially in Australia since 2003/ 04 (Fitt, 2003). These Bt proteins have highly specific insecticidal activity against caterpillar pests of cotton, including the target species in Australia Helicoverpa armigera (Hübner) and Helicoverpa punctigera (Wallengren). The two Bt proteins provide high insecticidal activity throughout the season against Helicoverpa spp. However, while most larvae feed and die prior to reaching second instar, there have been reports since 2005 of larvae surviving in up to 18% of the area planted to Bollgard II cotton (Whitburn and Downes, 2009). Larvae have reached or exceeded economic threshold levels on Bollgard II cotton (two larvae >3 mm per m of row over two consecutive checks, or one larva >8 mm per m row)

* Corresponding author. Tel.: þ61 7 4634 8300; fax: þ61 7 4634 8500. E-mail address: [email protected] (K. Knight). 0261-2194/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.cropro.2012.10.014

when there have been high egg lays (80e100 eggs/m) but also have occurred where egg lays have been low (4e10 eggs/m) (K. Knight, unpublished data). Since these reports of survivors in Bollgard II, collections of these larvae have been made and tested with F1 and F2 screening protocols (Andow and Alstad, 1998; Gould et al., 1997). It was found that larvae surviving in Bollgard II are no more likely to be carrying resistance alleles to the proteins in Bollgard II than eggs collected from other hosts (Downes, 2011). What remains unclear is how and why susceptible larvae able to survive occasionally on Bollgard II cotton. Barber (2008) suggested that the survival must be due to a reduction in the expression of the Cry1Ac and/or the Cry2Ab proteins but to date there has not been adequate data to assess this possibility. Many studies (reviewed by Dong and Li, 2007) have indicated that the level of Cry1Ac protein in Bollgard (Ingard in Australia) and Bollgard II cotton plants declines over the course of the growing season (Adamczyk et al., 2001; Greenplate, 1999; Holt, 1998; Olsen et al., 2005) while the Cry2Ab protein has been reported to remain at relatively constant levels (Adamczyk et al., 2001). In addition, various environmental factors can affect the

K. Knight et al. / Crop Protection 44 (2013) 50e58

level of the two Cry proteins in Bollgard II cotton plants. Luo et al. (2008) found that a combination of waterlogging and salinity decreased the expression of Cry1Ac by 45e72%, while waterlogging alone caused a decrease in expression of between 38 and 50%. Apparently as a consequence, waterlogging was found to significantly affect the control of neonates (Luo et al., 2008). Extended periods of high temperatures (>37  C) at boll setting also were found to decrease the expression of Cry1Ac in Bollgard II plants (Chen et al., 2005). Olsen et al. (2005) found that in pre-square plants there were significant changes in Cry1Ac expression, with warmer conditions enhancing control of Helicoverpa spp. larvae and cooler conditions decreasing larval mortality. Similarly, the expression of Cry2Ab in Bollgard II was reduced by low temperatures (<14  C) during the flowering/fruiting growth period for up to six days after the initial stress event (Addison and Rogers, 2010). Lu et al. (2011) examined the preferences of H. armigera larvae for various plant parts on Bollgard II and non-transgenic cotton and found that neonates preferred flowers to other plant parts on Bollgard II cotton more strongly than on non-Bt plants, pointing to the potential importance of neonate larval behaviour on survival outcomes on Bollgard II cotton. Collectively, the results of these earlier studies suggest that the low levels of observed survival of Helicoverpa larvae could be due to some combination of (a) the decline in Cry1Ac over the course of the growing season, (b) environmental factors affecting the expression of both Cry1Ac and Cry2Ab2, or (c) behavioural resistance (Lu et al., 2011). The purpose of the current research was to evaluate season-long expression of both of the Bt proteins in Bollgard II cotton to determine if fluctuations in protein expression could be a cause of the observed low levels of survival of Helicoverpa larvae on Bollgard II cotton, as recommended by Lu et al. (2011). The following specific hypotheses were investigated: 1. Does Bt protein expression vary with the position in the field, e.g. head ditch versus tail drain? 2. Does Bt expression vary among fields/farms within a production region? 3. Does Bt expression vary among Bt cotton cultivars? 4. Does the change in Bt protein expression over the season vary among different plant parts? In a separate study, bioassays with both Helicoverpa spp. were conducted on a subset of the samples reported here to determine if the observed expression levels were affecting control of the target Lepidoptera (Knight et al., unpublished data). Together these two sets of data provide the baseline to better understand and evaluate causes of the occasional presence of live Helicoverpa larvae in Australian Bollgard II cotton. 2. Materials and methods 2.1. Study overview 2.1.1. Trial locations The experiments were conducted in the cotton growing seasons of 2007/08, 2008/09 and 2009/10 in four cotton growing areas of Queensland, Australia; St George, Goondiwindi, Emerald and the Burdekin. In the Burdekin e a winter production area e two commercial varieties of Bollgard II, 289B and 60BRF, were planted on 29 January 2008 at the Ayr Queensland Department of Primary Industries and Fisheries Research Station. In 2008/09, the trial in Goondiwindi was planted on a commercial farm with two varieties 71B and 412B. In 2009/10, the trials were run on three farms at St George with the variety 71BRF. The Emerald trial in 2009/10 involved two fields on the same commercial farm; one of the fields

51

had a history of larvae surviving (Field 1) while the other field had no history of survivors (Field 2). The variety planted in Emerald was 71BRF. The farmers’ commercial crop-scouting services found no unusual survival of larvae in any of these fields during our sampling periods, indicating that Bollgard II was fully effective in all these fields at the levels at which the Cry1Ac and Cry2Ab proteins were expressed. 2.1.2. Tissue collections For all experiments, the tissues sampled were the first unfurled leaf, and first-position squares, small bolls and large bolls. In 2007/ 08, in the Burdekin collections were taken weekly from first flower (1 flower per metre) for eight weeks. In 2008/09 and 2009/10, St George, Goondiwindi and Emerald collections were taken fortnightly from early squaring (1 square per metre), weekly through the flowering period and fortnightly again until first defoliation. Once collected, the tissue was placed into pre-labelled zip lock plastic bags and placed directly into a cooler with an ice brick. For the Burdekin experiment, the tissue was sent to the Monsanto Research Facility in Toowoomba, Queensland, Australia by overnight courier. For the St George trials the material was transported to the Monsanto Research facility in a car fridge with a temperature of approximately 5  C. In Goondiwindi and Emerald, the material was processed on site. All tissue was frozen within 24 h of collection. 2.1.3. Within-field sampling At all experimental sites, fixed points for sampling throughout the season were determined at the first sampling date. The positions were at least 30 m from the head ditch and the tail drain, and in the middle of the field. At each sampling date, 20 of each of the plant structures were collected within 2 rows on either side of the sampling point. 2.1.4. Quantification of Cry protein expression using enzyme-linked immunosorbent assays (ELISA) Tissues were frozen (20  C) for at least 24 h, freeze-dried and finely ground. Three replicates of 5e10 mg of each tissue, from each field-position, from each sample date, from each trial location were sent to Monsanto Company, St Louis, USA for ELISA analysis. The Bollgard II tissues were evaluated for Cry1Ac and Cry2Ab2 expression by using toxin-specific double-antibody sandwich ELISA procedures as described by Sims et al. (1996) and as modified by Greenplate et al. (2003). 2.2. Statistical analysis All statistical analysis was performed in GenStat V13 (Payne, 2010). For all data sets, the results of the three laboratory replicates from each tissue sample were averaged prior to statistical analysis because the variation of interest is between different times, varieties, and farms at the sample level and not between subsamples derived from a single field sample. That is, the variation between sub-samples measures only laboratory measurement error. 2.2.1. Within-field variation of Cry protein levels Under the furrow-irrigation system that is standard practice in the Australian cotton industry, there can be differences in moisture regimes between the top, the middle, and close to the tail drain at the bottom of each field. This analysis aimed to evaluate whether field position was a significant source of variation in the Cry protein levels, compared with the other sources of variation in the sampling design. Because of substantial seasonal non-overlap between data sets for the different plant parts in crops planted at

52

K. Knight et al. / Crop Protection 44 (2013) 50e58

the standard planting time, each plant part was analysed separately. For each dataset, a repeated-measures restricted maximum likelihood (REML) analysis was performed fitting all main effects and all two-way interactions involving field position but not fitting any three-way or higher interactions, or interactions involving only factors other than field position. This analysis is analogous to analysis of an unreplicated factorial, as described by Cochran and Cox (1957, pp. 218e219), and provides a test of the magnitude of field-position effects relative to other variation in the data. Residual and normal plots were used to decide if data transformation was required prior to analysis. 2.2.2. Temporal, varietal and spatial variation of Cry protein levels Because all the field position and field-position interaction terms in the preliminary REML analyses were non-significant (see Results), field position was fitted in the random model of REML analyses that assessed seasonal, varietal and locality effects on Cry protein levels. Residual and normal plots were used to decide if data transformation was required prior to analysis. Where terms in the REML table of fixed effects were significant and more detail was required of significant differences, multiple-comparison tests were performed using t-tests with a sequential Bonferroni correction to maintain family-wise probability levels (Quinn and Keough, 2002 pp. 49e50). 3. Results

sets reported here, the field position factor was regarded as a replication factor and included in the random model for all subsequent analyses of these data sets. 3.2. Intra-regional and temporal effects on Cry protein content 3.2.1. Cry1Ac protein analyses 3.2.1.1. St George. For all four plant parts, the sampling-date  farm interaction was not significant (Table 2), and consequently both sampling-date and farm effects were considered at the main effect level only (Fig. 2A, Table 3). Average Cry1Ac content only varied between farms for large bolls; for the other three plant parts, Cry1Ac protein content was equivalent on all farms. This result points to stability of Cry1Ac expression across farms within a production region. However, there was marked temporal variation in Cry1Ac content of leaves at St George in 2009/10 (Fig. 2A), with a fall in protein level from the start of sampling in November 2009 to the beginning of 2010. After that, Cry1Ac content of leaves remained approximately constant through to the end of sampling. There were also significant differences between dates for small and large bolls, but this variation was relatively small, compared to the variation in leaf protein (Fig. 2A). The Cry1Ac content of squares did not differ significantly over the sampling period (Table 2). Up to the end of 2009, the level of Cry1Ac protein in leaves was markedly higher than in the other three plant parts, but as the end of the season approached, content of all plant parts converged to approximately 4 mg/g.

3.1. Field-position effects For each Cry protein, a total of 19 data sets were analysed from four Australian cotton-growing regions over three years and four plant parts giving a total of 114 separate estimates of the significance of field-position and interactions involving field-position (Table 1). Not one was significant at P ¼ 0.05 and two-thirds of the probabilities were 0.50 or larger (Table 1). Very clearly, there is not significant systematic variation in the content of either Cry proteins within fields with respect to sampling position from the top to the bottom of the field. Table 1 also give typical data (St George, 2009/10) to provide a more detailed picture of the lack of this variation. The seasonal trends in Cry protein levels at St George in 2009/10 are plotted in Fig. 1 along with the standard errors of difference (SEDs) for each sampling date. Any differences between field positions have to be approximately twice the plotted SED to be significant even for a simple LSD test (and before any Bonferroni probability correction, as used here). From inspection of Fig. 1, it is clear that differences between field positions within a field are inconsequential, compared to other sources of variation. Because of the absence of differences between field positions in any of the data

3.2.1.2. Emerald. On the Emerald farm, the sampling date  field interaction was not significant for all four plant parts (Table 4), and so both sampling-date (Fig. 3A) and field effects were considered at the main-effect level only. As for the St George data, average Cry1Ac content only varied between fields for large bolls with Field 1 averaging 5.2 mg/g and Field 2 4.8 mg/g; for the other three plant parts, Cry1Ac protein content was equivalent on both fields. This again points to stability of Cry1Ac expression across locations within a production region. There was significant variation in Cry1Ac content of leaves at Emerald from sampling date to sampling date in 2009/10 but with little, if any, overall seasonal trend (Fig. 3A). There were significant differences between dates for small bolls, but as for leaves this variation did not show any obvious seasonal trend. The Cry1Ac content of squares and large bolls did not differ significantly over the sampling period (Fig. 3A). In the case of squares, the REML table of fixed effects (Table 4) indicated a difference but no individual comparisons were significant once the Bonferroni probability correction was applied (Fig. 3A). At Emerald, the level of Cry1Ac protein in all plant parts was relatively stable over the course of the

Table 1 Summary of probabilities in the tables of fixed effects associated with field position and interactions involving field position in preliminary restricted maximum likelihood analyses of Cry1Ac and Cry2Ab protein content, plus detailed results from St George, 2009/10. Fixed term in REML analysis

Cry1Ac Field position Field position*date Field position*variety Field position*farm Cry2Ab Field position Field position*date Field position*variety Field position*farm

Overall summary

St George 2009/10

Average probability (and range)

Number of analyses

Leaf

Square

Small boll

Large boll

0.569 0.720 0.650 0.595

(0.118e0.943) (0.283e0.997) (0.466e0.745) (0.136e0.984)

19 19 7 12

0.802 0.962 e 0.326

0.291 0.941 e 0.308

0.335 0.409 e 0.935

0.840 0.533 e 0.984

0.520 0.715 0.641 0.617

(0.055e0.989) (0.313e0.999) (0.286e0.896) (0.135e0.936)

19 19 7 12

0.903 0.970 e 0.904

0.309 0.407 e 0.634

0.558 0.602 e 0.351

0.809 0.381 e 0.764

K. Knight et al. / Crop Protection 44 (2013) 50e58

53

Fig. 1. Variation in Cry1Ac protein expression level over field position and time for (A) leaves and squares, (B) small and large bolls, and Cry2Ab expression levels for (C) leaves and squares, (D) small and large bolls, St George 2009/10. Error bars are the average standard errors of difference for each sampling date.

season, with the content of all plant parts being in the range of 4e 8 mg/g for most of the season. 3.2.1.3. Summary. In these two sets of Cry1Ac data there was relatively little intra-regional variation in Cry1Ac expression in Bollgard II cotton. Additionally, Cry1Ac content for all plant parts was in the 4e6 mg/g range season-long, except for leaves at St George which started the season at w12 mg/g before falling to the 4e6 mg/g range by mid-season. 3.2.2. Cry2Ab protein analyses 3.2.2.1. St George. As for Cry1Ac, the sampling date  farm interaction was not significant (Table 2) for any plant part, meaning that

both sampling date and farm effects were considered at the main effect level only (Fig. 2B, Table 3). Average Cry2Ab content varied between farms for leaves and large bolls; for the other plant parts, Cry2Ab protein content was equivalent on all farms (Table 3). For leaves, Farm 3 was the lowest, and for large bolls it was equal lowest (Table 3). There was substantial early-season variation in the Cry2Ab content of leaves, again with a seasonal decline, but Cry2Ab content remained between 300 and 400 mg/g from the beginning of January to the end of the season (Fig. 2B). Squares also showed this gradual seasonal decline, but with protein levels remaining approximately constant at w800 mg/g after the beginning of January. Small-boll content varied somewhat erratically over time, but overall

Table 2 Tables of fixed effects for restricted maximum likelihood analyses of Cry1Ac and Cry2Ab protein content, St George, 2009/10. Fixed effect

Leaf df

Cry1Ac Sampling date Farm Date  farm Cry2Ab Sampling date Farm Date  farm

Square F test

Small boll

Prob.

df

F test

Prob.

9, 7.1 2, 15.7 18, 7.5

64.95 8.08 2.91

<0.001 0.004 0.069

5, 3.3 2, 7.6 10, 3.7

2.82 0.17 1.68

0.198 0.844 0.336

9, 3.3 2, 16.2 18, 1.8

26.63 104.44 22.27

0.008 <0.001 0.054

5, 34 3, 34 10, 34

9.29 1.29 1.97

<0.001 0.287 0.070

Large boll

df

F test

Prob.

df

F test

Prob.

4, 28 2, 28 8, 28

7.53 0.90 2.16

<0.001 0.420 0.063

4, 3.6 2, 5.6 8, 3.9

7.47 24.54 0.81

0.048 0.002 0.632

4, 12.5 2, 24 8, 13.8

7.51 1.98 1.26

0.003 0.159 0.337

4, 6 2, 11.6 8, 6.7

40.12 5.50 0.83

<0.001 0.021 0.604

54

K. Knight et al. / Crop Protection 44 (2013) 50e58

tested statistically as there were a limited number of dates on which all four plant parts were sampled. 3.2.2.2. Emerald. As for the other intra-regional data sets, the sampling-date  field interaction was not significant (Table 4) for any plant part; again both sampling date (Fig. 3B) and field effects were considered at the main-effect level only. Average Cry2Ab content varied between fields for leaves with Field 1 averaging 511.2 mg/g, and Field 2 488.9 mg/g but for the other plant parts, Cry2Ab protein content was equivalent on the two fields. There was a seasonal decline in the Cry2Ab content of leaves at Emerald (Fig. 3B), but Cry2Ab content remained approximately constant at w400 mg/g from the beginning of January to the end of the season. Cry2Ab content of squares remained constant throughout the season at w800 mg/g. Small-boll content varied somewhat erratically over time, but remained in the range 500e 900 mg/g over the sampling period. As for St George, there was a rising trend in Cry2Ab content of large bolls, but this was not significant at Emerald. Overall, the level of Cry2Ab protein was markedly higher in square and small bolls than in large bolls (Fig. 3B), as was the case at St George. 3.2.2.3. Summary. Cry2Ab expression showed some intra-regional variation, but this was primarily in the leaves. In both regions, there was no significant intra-regional variation in Cry2Ab content of squares or small bolls. Temporal changes in expression were also detected, with the Cry2Ab content of leaves falling during the season while the large-boll Cry2Ab content rose. 3.3. Cultivar and temporal effects on Cry protein content

Fig. 2. Temporal and plant-part variation in (A) Cry1Ac and (B) Cry2Ab protein content, St George, 2009/10. For each protein and plant part, dates followed by the same letter are not significantly different (t-test with a sequential Bonferroni correction (Family-wise probability P ¼ 0.05)). Where only a single pair of dates was different, the two points are marked with *.

remained approximately constant over the sampling period. In contrast, there was a rising trend in Cry2Ab content of large bolls. Overall, the level of Cry2Ab protein was markedly higher in squares and small bolls than leaves and large bolls (Fig. 2B) but this was not

Table 3 Season-long mean Cry1Ac and Cry2Ab protein content (mg/g) of leaves, squares, small and large bolls from three farms at St George, 2009/10. Parameter Cry1Ac Leafa Squareb Small bollb Large bolla Cry2Ab Leafa Squareb Small bollb Large bolla

Farm 1 6.960 a 3.481 4.625 3.836 ab 616.9 a 937.3 1083.0 349.7 b

Farm 2 6.618 a 3.517 4.760 3.968 a 560.4 b 963.0 1174.0 405.7 a

Farm 3 6.079 a 3.392 4.906 3.472 b 462.4 c 892.5 1034.0 358.6 b

a REML F-test significant so means compared using t-tests with a sequential Bonferroni correction (Family-wise probability P ¼ 0.05). Within Cry proteins and plant parts, means followed by the same letter are not significantly different. b REML F-test not significant so no means testing performed.

3.3.1. Cry1Ac protein analyses 3.3.1.1. Burdekin. For all three plant parts, there were no interactions between sampling date and cultivar, and no differences between sampling dates for either square or small boll (Fig. 4A) (all P > 0.05 in the REML table of fixed effects, data not shown but patterns are similar to those in Tables 2 and 4). The only differences between dates in Cry1Ac leaf content was between samples collected on 12 and 21 May 2008 (Fig. 4A). In summary, there was relatively little variation over time in Cry1Ac levels at this site. The two cultivars, 289B and 60BRF had equivalent Cry1Ac contents in both squares and leaves (P > 0.05), but for small bolls 60BRF (4.76 mg/g) had a higher Cry1Ac season-long average than did 289B (3.82 mg/g) (P ¼ 0.001). 3.3.1.2. Goondiwindi. There were significant differences in Cry1Ac content between cultivars for all four plant parts and significant sampling date  cultivar interactions for three of them (Table 5). The overall picture was for cultivar 412B to have higher levels of Cry1Ac protein than cultivar 71B (Fig. 5A and B). Where both a cultivar main-effect and an interaction existed, the focus of multiple testing was on week-by-week cultivar comparisons (Fig. 5A and B). For leaves, the cultivar 412B had a significantly higher Cry1Ac content than cultivar 71B for all sampling dates except the final two, while for squares, 412B was significantly higher than 71B for all dates except 19/1/09. For small bolls, there was no interaction (P ¼ 0.084) so testing was done at the main effect level only; cultivar 412B (5.78 mg/g) had a significantly higher season average than 71B (4.24 mg/g). For large bolls, the two cultivars were significantly different only on 2 February 2009, again with 412B being the higher. 3.3.1.3. Summary. At these two sites, within broadly similar seasonal patterns there were differences both in the overall expression Cry1Ac expression level between cultivars and in the

K. Knight et al. / Crop Protection 44 (2013) 50e58

55

Table 4 Tables of fixed effects for restricted maximum likelihood analyses of Cry1Ac and Cry2Ab protein content, Emerald, 2009/10. Fixed effect

Leaf df

Cry1Ac Sampling date Field Date  field Cry2Ab Sampling date Field Date  field

8, 9.2 1, 26.6 8, 9.2 8, 8 1, 3 8, 8

Square

Small boll

F test

Prob.

df

F test

Prob.

df

5.68 2.80 2.62

0.008 0.106 0.084

8, 26.6 1, 11.4 8, 26.6

3.26 0.25 1.26

0.010 0.626 0.306

5, 20 1, 2 5, 20

9.31 23.12 1.93

0.002 0.015 0.186

8, 6.3 1, 9.3 8, 6.3

0.38 3.26 1.31

0.898 0.103 0.376

5, 17.9 1, 9.2 5, 17.9

changes over time for some plant parts. This suggests that Bollgard II cotton cultivars can exhibit somewhat different expression profiles, but within broadly similar seasonal expression patterns.

Large boll F test

Prob.

df

F test

Prob.

5.34 1.56 0.82

0.003 0.338 0.548

5, 3 1, 3.8 5, 3

3.47 117.89 3.58

0.168 <0.001 0.163

11.41 0.76 1.01

<0.001 0.406 0.438

5, 2.9 1, 7.9 5, 2.9

3.63 1.39 2.55

0.167 0.272 0.244

3.3.2. Cry2Ab protein analyses 3.3.2.1. Burdekin. For leaves and small bolls, there were no interactions between sampling date and cultivar, but a significant cultivar  date interaction existed for squares (P ¼ 0.011). Examination of this interaction indicated that while the overall seasonal pattern was similar, peak protein content occurred later in cultivar 60BRF

than in 289B, leading to a significant difference on one date (Fig. 4B). However, the overall seasonal pattern of Cry2Ab content was essentially the same for all plant parts, with a slight declining trend in content over time. This trend was statistically significant only for leaves, in that the Cry2Ab content on the final sampling date was lower than on two dates at the beginning of the season (Fig. 4B). Cultivars were not different at the main effect level for leaves and squares (P < 0.05), but for small bolls Cry2Ab content of 60BRF (1275 mg/g) was markedly higher than for 289B (768 mg/g) (P < 0.001).

Fig. 3. Temporal and plant-part variation in (A) Cry1Ac and (B) Cry2Ab protein content, Emerald, 2009/10. For each protein and plant part, dates followed by the same letter are not significantly different (t-test with a sequential Bonferroni correction (Family-wise probability P ¼ 0.05)).

Fig. 4. Temporal, cultivar and plant-part variation in (A) Cry1Ac and (B) Cry2Ab protein content, Burdekin, 2007/08. For each protein and plant part, dates followed by the same letter are not significantly different (t-test with a sequential Bonferroni correction (Family-wise probability P ¼ 0.05)). For Cry2Ab content of squares, the two cultivars were significantly different on dates marked *.

56

K. Knight et al. / Crop Protection 44 (2013) 50e58

Table 5 Tables of fixed effects for restricted maximum likelihood analyses of Cry1Ac and Cry2Ab protein content, Goondiwindi, 2008/09. Fixed effect

Cry1Ac Sampling date Cultivar Date  cultivar Cry2Ab Sampling date Cultivar Date  cultivar

Leaf

Square

Small boll

Large boll

df

F test

Prob.

df

F test

Prob.

df

F test

Prob.

df

F test

Prob.

11, 11.4 1, 21.5 11, 11.4

35.36 424.77 8.46

<0.001 <0.001 <0.001

8, 8 1, 6.6 8, 8

67.84 488.52 7.36

<0.001 <0.001 0.005

7, 7.2 1, 19.4 7, 7.2

5.90 133.14 2.97

0.015 <0.001 0.084

5, 5.3 1, 4 5, 5.3

15.86 7.97 30.83

0.003 0.048 <0.001

11, 10.8 1, 22.6 11, 10.8

43.75 15.46 6.56

<0.001 <0.001 0.002

8, 8 1, 4.5 8, 8

31.59 49.47 6.31

<0.001 0.001 0.009

7, 8.5 1, 12.9 7, 8.5

9.47 74.94 6.79

0.002 <0.001 0.006

5, 22 1, 22 5, 22

28.41 6.58 2.34

<0.001 0.018 0.076

3.3.2.2. Goondiwindi. The Cry2Ab leaf data required transformation onto the log10(x þ 1) scale prior to analysis; equivalent means are presented in Fig. 5C. As for Cry1Ac content, there were significant differences in Cry2Ab content between cultivars for all four plant parts and significant sampling date  cultivar interactions for three of them (Table 5). However, the overall picture of cultivar differences was much less clear cut than for Cry1Ac, but with similar seasonal patterns in both cultivars (Fig. 5C and D). For leaves, 412B was higher than 71B for 10 of the 12 sampling dates, but date-by-date testing showed significant differences on only two

dates, 19 and 27 January 2009 with one difference in each direction (Fig. 5C). Over the whole season, the average level in 412B leaves (1015 mg/g) was only 10.6% higher than in 71B leaves (913 mg/g). Cry2Ab content of squares on 412B was higher than on 71B on seven of nine dates but there was only three dates with significant differences (Fig. 5C). On 27 January and 2 March, Cry2Ab levels in 412B squares were higher than in 71B squares but on 27 January the reverse was true (Fig. 5C). Cry2Ab content of small bolls fluctuated markedly from week-to-week and on two sampling dates, content of 412B small bolls was significantly higher than in 71B bolls. There

Fig. 5. Temporal and cultivar variation in Cry1Ac protein content in (A) leaf and square and (B) small and large bolls, and Cry2Ab protein content in (C) leaf and square, and (D) small and large bolls at Goondiwindi, 2008/09. For each plant part and Cry protein, cultivars significantly different for each date marked are * (t-test with a sequential Bonferroni correction (Family-wise probability P ¼ 0.05)). Where no cultivar differences exist for each Cry protein and plant part, sampling dates with the same letters are not significantly different (t-test with a sequential Bonferroni correction (Family-wise probability P ¼ 0.05)).

K. Knight et al. / Crop Protection 44 (2013) 50e58

was no cultivar  date interaction for large bolls (P ¼ 0.076) and the Cry2Ab content of cultivar 412B (759.8 mg/g) was higher than in 71B (629.1 mg/g) (P ¼ 0.018). Over the course of the season, Cry2Ab content of large bolls increased significantly (Fig. 5D). 3.3.2.3. Summary. As for Cry1Ac, there were some cultivar and cultivar  time differences in Cry2Ab expression for some plant parts, but with broad overall similarity across cultivars. 4. Discussion Diagnostic concentrations for resistance monitoring in Helicoverpa spp. have been established as 0.25 mg/cm2 and 2.0 mg/cm2 for Cry1Ac and Cry2Ab, respectively, in diet surface-contamination bioassays (Mahon et al., 2007). Using the range of published data on specific leaf weight and area (see Supplementary Table 1), we expressed the range of leaf Cry protein concentrations reported here (Table 3, Figs. 2e4) in the same units, i.e. mg/cm2 of leaf. This pointed to leaf Cry1Ac concentrations being potentially in the range 0.02e0.12 mg/cm2 and Cry2Ab concentration between 1.56 and 9.31 mg/cm2. Despite the operational differences between Helicoverpa larvae feeding on leaves and on insect diet, the congruence of these extrapolations and the discriminating doses supports the field observations that in all of the commercial fields sampled for this study, Bollgard II was fully effective. Furthermore, the results demonstrate that, over a series of years, locations and cultivars, there was not biologically significant variation in the expression level of either of the Cry proteins within Bollgard II fields. For Cry1Ac, protein expression also was relatively uniform among fields on a farm and among farms in a region for all plant parts tested. Large bolls were the plant parts that showed the most variation in Cry1Ac protein content among farms and fields. For Cry2Ab, protein expression was uniform among fields on a farm, and among farms in a region for both squares and small bolls. For leaves, there was variation between fields on a farm and among farms within a region, while for large bolls the significant variation was between farms in a region but not between fields on a farm. In the case of large bolls, there was a rising trend for the Cry2Ab protein over time. However, the results confirm that there are temporal changes in both Cry proteins over the season and spatial differences in Cry protein content among different plant structures (Adamczyk et al., 2001; Kranthi et al., 2005; Olsen et al., 2005). Although there was an initial e and often significant e decline in both proteins, especially in leaves, this did not continue for the entire season. Typically, protein content remained approximately stable from 19 nodes, or approximately 7 NAWF (nodes above white flower), onwards. In the case of large bolls, Cry2Ab protein content even increased later in the season. Expression of Cry1Ac in leaves was higher than in the other plant parts tested, though by the end of the season Cry1Ac content was similar for all plant structures. The trend for Cry2Ab was that squares and small bolls had higher expression levels than leaves and large bolls. In addition, the data demonstrate that there can be differences among cultivars in Bt protein expression, but not for the entire season or for all plant parts. In the Burdekin, there were only differences among cultivars in Cry1Ac and Cry2Ab content in small bolls whereas in Goondiwindi the Cry1Ac and Cry2Ab content were different for all the plant parts tested. However, these differences among cultivars were small, as would be expected given that new cultivars are tested to ensure that the level of expression of both Cry1Ac and Cry2Ab is not significantly less than an established reference variety prior to commercialisation. Collectively, the results demonstrate that the content of Cry1Ac and Cry2Ab in Bollgard II plants is stable within a field, among

57

fields, and among farms, while there is some variation in protein content among plant parts and a trend of declining production of both the Cry proteins over the season. Given the pattern of larval survival observed with Bollgard II, with survival varying among field and farms, the data presented here indicate that overall patterns of Cry protein expression cannot be the only variable that allows larvae to survive Bollgard II; other biotic factors must contribute to this phenomenon perhaps through their interactions with Cry protein expression. These potential factors include (a) physiological stress on the plant, (b) variation in susceptibility in the pest population, and (c) larval behaviour. The Bt proteins in transgenic cotton plants are a component of the total soluble proteins present. Luo et al. (2008) suggested that reduction in the levels of Bt protein in transgenic cotton could simply be part of the overall levels of soluble protein in cotton caused by environmental stresses. In their experiment on waterlogging and salinity stress on Cry1Ac cotton, Luo et al. (2008) found that the combination of both these stresses inhibited nitrogen metabolism which reduced production of total protein. In contrast, we found no differences in Cry expression between the top and potentially-waterlogged tail-drain sections of fields. High temperatures also degraded total soluble protein by 47e55% during the boll setting period when it was found that insecticidal activity decreased in Cry1Ac expressing cotton (Chen et al., 2005). This indicates that abiotic factors can affect the total soluble proteins present in Bollgard II cotton plants and hence the levels of Bt proteins, either transiently or throughout the season, but none of these effects were evident in our studies. It is also well known that there is wide variability of susceptibility to the Cry proteins in Helicoverpa spp. populations (Adamczyk et al., 1998; Greenplate et al., 1998; Stone and Sims, 1993) and this also has the potential to impact survival on Bollgard II cotton observed in the field. In Australia, Liao et al. (2002) found 3 strains of H. armigera that varied significantly in their susceptibility to Cry1Ac and that both H. armigera and H. punctigera are more tolerant of a range of insecticidal proteins than Heliothis virescens. H. armigera field populations tested by Bird and Akhurst (2007) had a 4.6 and a 6.6-fold range in susceptibility to Cry1Ac and Cry2Ab respectively. For H. punctigera the range was 3.2 and 3.5-fold (Bird and Akhurst, 2007). However, tests of H. armigera survivors from Bollgard II cotton found no differences in the frequency of either resistance gene between random samples and survivors from Bollgard II cotton (Downes, 2011). Thirdly, Helicoverpa spp. also have the ability to avoid the Cry toxins in its diet. In a choice situation, H. armigera strongly avoided meridic diet that contained Cry1Ac compared to standard diet, and also preferred feeding on meridic diet to Cry1Ac plant material (Singh et al., 2008). Gore et al. (2005) found that both Helicoverpa zea and H. virescens selected diet with low concentrations of Cry1Ac over diet with higher concentrations of the toxin, while for Cry2Ab, the avoidance was not as clear. In a study by Singh et al. (2008), H. armigera was also able to select diet with Cry1Ac concentrations at sublethal doses, which may mean that larvae have an increased chance of survival on transgenic cotton plants through such behavioural responses. Behaviour of larvae on Bt cotton has been shown to differ from larvae on non-Bt cotton. Lu et al. (2011) showed that in choice tests involving leaves, flowers, squares and small bolls, neonates showed a stronger preference for flowers in tests using Bollgard II cotton than in tests using non-Bt cotton. They also found no differences in preferences between plant parts between larvae that were the progeny of survivors from Bollgard II or from non-Bt hosts. In addition, Gore et al. (2002) found that larvae move twice as far down the plant on Bt cotton compared with non-Bt cotton. On Bt cotton, fewer larvae remained in the terminals and squares than

58

K. Knight et al. / Crop Protection 44 (2013) 50e58

non-Bt cotton and larvae began migrating within 3 h, and by 6 h after infestation had moved more than 4 nodes below the plant terminal (Gore et al., 2002). Gore et al. (2002) showed that a higher percentage of larvae were lower in the plant canopy within flowers and bolls in Bt cotton than in non-Bt cotton. Because cotton begins flowering from the bottom of the plant and younger fruit forms at the top of plant, as larvae move down the plant they are likely to encounter flowers and as they continue to forage down they are likely to feed on bolls. The lower expression of Cry protein in white flowers may be such that larvae are capable of overcoming the adverse effects of Cry1Ac and Cry2Ab toxicity (Gore et al., 2001). The increased movement of larvae on Bt cotton could provide greater opportunity for encountering white flowers, thus increasing the chance of surviving Bollgard II cotton. Taken together, these studies indicate that Helicoverpa species have a diversity of behavioural responses that could contribute to the occurrence of larger larvae on Bollgard II cotton. Overall, this study shows that the expression of the Cry proteins in Bollgard II is relatively uniform within and among cotton fields and farms, though the data confirm that there are differences in protein expression among plant structures and varieties throughout the season. Taking this into account, when larvae are found to be surviving in Bollgard II, factors other than protein expression need to be considered, probably acting in concert with protein expression. Biotic or abiotic factors may have caused a decline or fluctuation in the total soluble proteins and hence the Cry proteins in Bollgard II plants. Larval susceptibility as well as the behaviour of increased movement on Bollgard II could give larvae an opportunity to encounter white flowers that have lower expression and high nutritional value. The ability to avoid Cry proteins also gives larvae an increased opportunity to survive and develop on Bollgard II. We have shown that generally Bollgard II protein production is stable and that other factors need to be considered when there are larvae ‘escapes’. Acknowledgements Christie Warburton and Riedha Ekalianna provided excellent technical assistance. Jamie Street assisted with sampling in St George and the Geddes family took and processed all of the samples for the trials in Emerald. Dr. Sharon Downes and her team from CSIRO Narrabri assisted in sampling in St George. Sincere thanks to all of the growers who allowed us to work on their farms. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.cropro.2012.10.014. References Adamczyk, J., Adams, L.C., Hardee, D.D., 2001. Field efficacy and seasonal expression profiles for terminal leaves of single and double Bacillus thuringiensis toxin cotton genotypes. J. Econ. Entomol. 94, 1589e1593. Adamczyk, J.J., Holloway, J.W., Church, G.E., Leonard, B.R., Graves, J.B., 1998. Larval survival and development of the fall armyworm (Lepidoptera: Noctuidae) on normal and transgenic cotton expressing the Bacillus thuringiensis Cry1A(c) dendotoxin. J. Econ. Entomol. 91, 539e545. Andow, D.A., Alstad, D.N., 1998. F2 screen for rare resistance alleles. J. Econ. Entomol. 91, 572e578. Addison, S.J., Rogers, D.J., 2010. Potential impact of differential production of the Cry2Ab and Cry1Ac proteins in transgenic cotton in response to cold stress. J. Econ. Entomol. 103, 1206e1215. Barber, J., 2008. Larval survival in Bollgard II cotton. Aust. Cottongrower 29 (2), 10e13.

Bird, L.J., Akhurst, R.J., 2007. Variation in susceptibility of Helicoverpa armigera (Hübner) and Helicoverpa punctigera (Wallengren) (Lepidoptera: Noctuidae) in Australia to two Bacillus thuringiensis toxins. J. Econ. Entomol. 94, 84e94. Chen, D., Ye, G., Yang, C., Chen, Y., Wu, Y., 2005. The effect of high temperature on the insecticidal properties of Bt cotton. Environ. Exp. Bot. 53, 333e342. Cochran, W.G., Cox, G.M., 1957. Experimental Designs, second ed. Wiley, New York. Dong, H.Z., Li, W.J., 2007. Variability of endotoxin expression in Bt transgenic cotton. J. Agron. Crop Sci. 193, 21e29. Downes, S., 2011. End of Season Results from CSIRO Bt Resistance Monitoring 2010/ 11, Cotton Catchment Communities Report, 6 pp. Internet resource, available at: http://www.cottoncrc.org.au/industry/Publications/Pests_and_Beneficials (accessed 11.10.12.). Fitt, G., 2003. Deployment and impact of transgenic Bt cotton in Australia. In: Kalaitzandonakes, N. (Ed.), The Economic and Environmental Impacts of Agbiotech: A Global Perspective. Kluwer Academic/Plenum Publishers, New York, pp. 141e161. Gore, J., Leonard, B.R., Adamczyk, J.J., 2001. Bollworm (Lepidoptera: Noctuidae) survival on ‘Bollgard’ and ‘Bollgard II’ cotton flower bud and flower components. J. Econ. Entomol. 94, 1445e1451. Gore, J., Leonard, B.R., Church, G.E., Cook, D.R., 2002. Behavior of bollworm (Lepidoptera: Noctuidae) larvae on genetically engineered cotton. J. Econ. Entomol. 95, 763e769. Gore, J., Adamczyk, J.J., Blanco, C.A., 2005. Selective feeding of tobacco budworm and bollworm (Lepidoptera: Noctuidae) on meridic diet with different concentrations of Bacillus thuringiensis proteins. J. Econ. Entomol. 98, 88e94. Greenplate, J.T., 1999. Quantification of Bacillus thuringiensis insect control protein Cry1Ac over time in Bollgard cotton fruit and terminals. J. Econ. Entomol. 92, 1377e1383. Greenplate, J.T., Head, G.P., Penn, S.R., Kabuye, V.T., 1998. Factors potentially influencing the survival of Helicoverpa zea on Bollgard cotton. In: Proceedings, Beltwide Cotton Conference 1998. National Cotton Council of America, Memphis, TN, pp. 1030e1033. Greenplate, J.T., Mullins, J.W., Penn, S.R., Daham, A., Reich, B.J., Osborn, J.A., Rahn, P.R., Ruschke, L., Shappley, Z.W., 2003. Partial characterization of cotton plants expressing two toxin proteins from Bacillus thuringiensis: relative contribution, toxin interaction, and resistance management. J. Appl. Entomol. 127, 340e347. Gould, F., Anderson, A., Jones, A., Sumerford, D., Heckel, D.G., Lopez, J., Micinski, S., Leonard, R., Laster, M., 1997. Initial frequency of alleles for resistance to Bacillus thuringiensis toxins in field populations of Heliothis virescens. Proc. Natl. Acad. Sci. U. S. A. 94, 3519e3523. Holt, H.E., 1998. Season-long monitoring of transgenic cotton plants e development of an assay for the quantification of Bacillus thuringiensis insecticidal crystal protein. In: Proc. 9th Australian Cotton Conference. Australian Cotton Growers’ Research Association, Wee Waa, Australia, pp. 331e335. Kranthi, K.R., Naidu, S., Dhawad, C.S., Tatwawadi, A., Mate, K., Patil, E., Bharose, A.A., Behere, G.T., Wadaskar, R.M., Kranthi, S., 2005. Temporal and intra-plant variability of Cry1Ac expression in Bt-cotton and its influence on the survival of the cotton bollworm, Helicoverpa armigera (Hübner) (Noctuidae: Lepidoptera). Curr. Sci. 89, 291e298. Liao, C., Heckel, D.G., Akhurst, R., 2002. Toxicity of Bacillus thuringiensis insecticidal proteins for Helicoverpa armigera and Helicoverpa punctigera (Lepidoptera: Noctuidae), major pests of cotton. J. Invertebr. Pathol. 80, 55e63. Lu, B., Downes, S., Wilson, L., Gregg, P., Knight, K., Kauter, G., McCorkell, B., 2011. Preferences of field bollworm larvae for cotton plant structures: impact of Bt and history of survival on Bt crops. Entomol. Exp. Appl. 140, 17e27. Luo, Z., Dong, H., Li, W., Zhao, M., Zhu, Y., 2008. Individual and combined effects of salinity and waterlogging on Cry1Ac expression and insecticidal efficacy of Bt cotton. Crop Protect. 27, 1485e1490. Mahon, R.J., Olsen, K.M., Downes, S., Addison, S., 2007. Frequency of alleles conferring resistance to the Bt toxins Cry1Ac and Cry2Ab in Australian populations of Helicoverpa armigera (Lepidoptera: Noctuidae). J. Econ. Entomol. 100, 1844e1853. Olsen, K.M., Daly, J.C., Holt, H.E., Finnegan, E.J., 2005. Season-long variation in expression of Cry1Ac gene and efficacy of Bacillus thuringiensis toxin in transgenic cotton against Helicoverpa armigera (Lepidoptera: Noctuidae). J. Econ. Entomol. 98, 1007e1017. Payne, R.W., 2010. The Guide to GenStat Release 13. In: Statistics, vol. 2. VSNI, Hemel Hempstead, UK. Quinn, G.P., Keough, M.J., 2002. Experimental Design and Data Analysis for Biologists. Cambridge University Press, Cambridge. Singh, G., Rup, P.J., Koul, O., 2008. Selective feeding of Helicoverpa armigera (Hübner) and Spodoptera litura (Fabricius) on meridic diet with Bacillus thuringiensis toxins. J. Insect Behav. 21, 407e421. Sims, S.R., Pershing, J.C., Reich, B.J., 1996. Field evaluation of transgenic corn containing a Bacillus thuringiensis Berliner insecticidal protein gene against Helicoverpa zea (Lepidoptera: Noctuidae). J. Entomol. Sci. 31, 340e346. Stone, T.B., Sims, S.R., 1993. Geographic susceptibility of Heliothis virescens and Helicoverpa zea (Lepidoptera: Noctuidae) to Bacillus thuringiensis. J. Econ. Entomol. 86, 989e994. Whitburn, G., Downes, S., 2009. Surviving Helicoverpa larvae in Bollgard II: survey results. Aust. Cottongrower 30 (3), 12e16.