study finds that

inhibiting the

u ptake of nutrients

like manganese

mance. The “yield

GM Soybean: As can be seen from Table 2, no claims of dramatic yield increases can be made about the largest cultivated GM crop, the GM soybean. It is very clear that yield growth has come down in the years after GM soybean was introduced. In fact, the initial five years after the introduction of transgenic soy, which also witnessed the most rapid expansion of GM soy, saw a negative yield growth. In the subsequent years, yields seem to have recovered slightly but are still lower than the pre-GM years. The 2008 yields of US soybean (at 40 bushels per acre, as per the National Agricultural Statistics Service of the USDA), with 92% of such soybean being genetically modified, are lower than the 1994 yields of 41.4 bushels (before GM soybean was introduced). This is just to underline the fact that yields are much more complex in reality than a linear function of one technology.

Further, genetic engineering as a technology cannot increase yields per se since yields are a factor of complex, multi-genic traits, not in application with GM crops now. With GM soybean, decreased yields have been recorded even in field trials. A University of Nebraska study found that Roundup Ready (RR) soya varieties (Roundup Ready is the brand of Monsanto’s GM trait) yielded 5% less than their closest conventional relatives and 10% less than high-yielding conventional lines (Elmore et al 2001). This corresponds to a loss in production of nearly 200 kg/ha.

This yield drag of RR soy is also apparent in flat overall soybean yields from 1995 to 2003, the years during which GM soy adoption increases to 81% of US soybeanplanted land. A 2007 study by Kansas State University, led by Barney Gordon, an agronomist, suggests that RR soy continues to suffer from a “yield drag” (Gordon

2007). Gordon’s

drag” of RR soy has been correlated to the presence of glyphosate in the root zone of the plant.

GM Cotton: When it comes to cotton, a different picture than the one for soybean emerges. In the years prior to the introduction of transgenic cotton, the yield growth was 1.27 while in the post-1996 p eriod, it jumped to 2.30. However, what is important to note is that the 1996-2000 period saw a sharp decline in yield growths in cotton – this is the period when GM cotton spread to 61% of total cotton cultivation in the US. Therefore, it would be misleading to say that GM cotton has contributed to yield increases in cotton here. The subsequent years, which witnessed a slower expansion of GM cotton, saw an impressive increase in yield growth. Steep yield increases in cotton in 2004 and 2005 are in fact attributed to excellent weather conditions (Meyer et al 2007).

Going by the above data, it can be concluded that transgenic soybean or cotton did not result in yield growths and in fact, there is a clear trend of yield growth d eclines in the case of soybean after the a dvent of the transgenic, herbicide-tolerant soybean in the US. For both these crops in the US, the period of rapid expansion of transgenic crop cultivation (1996-2000) witnessed a negative yield growth c ompared to earlier years.

At the micro-level, a study by the USDA that looked at adoption of GM crops and its relation to net returns/yields throws some light on different GM crops (Fernandez-Cornejo et al 2002). The 2002 USDA study report says that the adoption of herbicidetolerant soybeans did not have a significant impact on net farm returns while the adoption of herbicide-tolerant corn improved farm net returns among specialised farms. The report goes on to add that the positive financial impact may be due to seed companies setting low premiums for herbicide tolerant corn in an attempt to expand market share. On Bt cotton, the report says there was a positive impact on net returns among cotton farms but adoption of Bt corn had a negative impact among specialised corn farms. The study raises a pertinent question to itself: “perhaps the biggest issue raised by these results is how to explain the rapid adoption of GE crops when financial impacts appear to be mixed or even negative”, it says suggesting that “other considerations may be motivating farmers” (what is now called the “convenience effect”).

The Chinese Picture

Coming to China, Xinjiang province is known for its cotton cultivation and around 30% to 32% of China’s cotton production is contributed by this one province. This province is supposed to have touched yields of

2.5 bales per acre of cotton by 2004. Between 1980 and 2004, the cotton yields in this province are reported to have increased by 3.74 times (Xiaoling et al 2006). What is interesting however is that a USDA Global Agriculture Information Network (GAIN) report of 2005 notes that in Xinjiang, Bt cotton varieties are not widely planted (“because bollworms are less of a problem there than in other areas, so farmers prefer seeds of the cheaper conventional varieties”) (GAIN 2005). This picture of planting is reinforced by other researchers (Huang et al 2002). When the overall adoption of Bt cotton within China’s cotton was 45% by 2001, Xinjiang gets classified within “rest of China” with its low 7% adoption rate here. Xinjiang’s high cotton yields are attributed to the planting of conventional varieties with specific traits, such as dwarf plant size and early maturity, as well as to new agronomic practices, including “high density

vol xliv no 22

INSIGHT

sowing, plastic sheet covering and drip i rrigation” (USDA 1 May 2007). The FoE (2008) report goes on to illustrate that the yields of Xinjiang are far higher than the rest of the country over the years that Bt cotton adoption has spread in China without such an adoption happening in this important cotton-growing province.

Bt Cotton in India

While this is the picture with China, it is interesting to note that the hype in India about Bt cotton hides important factors that have contributed to cotton yield

Table 4 enumerates the area, production and yields of cotton in the state of G ujarat, starting from 1997-98 as per the Cotton Corporation of India’s data. It has to be noted that these figures vary from the data put out by the Central Institute for Cotton Research (CICR) as well as the state department of agriculture, Gujarat (yield figures put out by state departments are based on “crop cutting experiments” and not on data from ginning factories).

In a letter dated 9 May 2005 (D O No IST/2003/Bt Cotton K-6 (PtII)), the agricultural secretary, government of Gujarat

as 2006 (Pandit 2008). Cotton production was estimated to have come down to 90 to 100 lakh bales from the earlier projected 125 lakh bales. Another report stated that cotton productivity in Gujarat, according to the directorate of agriculture, could have gone down by 78 kg/ha in 2006-07, coming down to 650 kg/ha from a record 728 kg in 2005-06 and could have fallen further in the 2007 kharif season (Trivedi 2008).

Incidentally, from 2002 onwards, the increase in irrigated cotton area in Gujarat has been steady. Table 5 is based on data obtained from the Directorate of Agriculture on the year-wise extents of irrigated and unirrigated cotton in Gujarat, in

i mprovements in the country. The Xinjiang province equivalent here seems to be G ujarat. Before looking at cotton in Gujarat, Table 3 gives the area, production and yield of c otton in India (1997-98 onwards).

Cotton is grown in several states of I ndia like Maharashtra, Gujarat, Andhra Pradesh, Punjab, Haryana, Tamil Nadu, etc, and the cotton cultivation in India is divided into southern, central and northern (cotton-growing) zones.

Gujarat is one of the most important cotton-growing states in India. By 2007, Gujarat’s share in total cotton production in India has gone up from around 29% at the time of introduction of Bt cotton to an impressive 39%. In fact, understanding the situation of cotton in Gujarat separately and not allowing it to distort the entire

Table 4: Cotton Area, Production and Yield in Gujarat

(1997-98 to 2006-07)

Source: Cotton Corporation of India’s web site.

wrote to the chairperson of the Genetic Engineering Approval Committee:

Yes, the productivity which was 175 kg/ha in 2002-03 is touching 460 kg/ha in 2004-05. But this is not solely due to Bt cotton hybrids as Gujarat recorded 450 in 1998-99 when there was no Bt cotton. In our opinion, all these years were good years with low to medium bollworm activity, hence this increase.

The agricultural secretary (government of Gujarat) is pointing to an important a spect related to yield analysis with insect resistant GM crops here – that if pest incidence itself is low due to climatic and

o ther conditions, there cannot be yield i ncreases due to protection from crop l osses through insect-resistant varieties! In India, it is recorded that bollworm incidence across regions has been quite low from 2001 (AICCIP Annual Reports).

Through their official monitoring and evaluation report of Bt cotton in 2006-07, the Gujarat state authorities have also r evealed that

the productivity of cotton crop also increased due to increase of irrigation facility by massive water harvesting programmes. On the other hand, rainfall is also very good during past three years. Parameters like irrigation facility, good monsoon, use of drip, low pest pressure, black soil and farmers’ experience are contributing in the success of cotton crop in the state.

Reports from the first quarter of 2008 indicate that cotton production in Gujarat declined in the kharif crop of 2007 as well Gujarat has been steadily increasing,

Table 5: Changes in Irrigated and Unirrigated Cotton in Gujarat (2000-01 to 2005-06)

Source: Right to Information obtained from the Department of Agriculture, Government of Gujarat, 2008.

b etween 2000-01 and 2005-06, there has been a 43.3% increase in the total area under irrigated cotton. Any analysis of cotton yield increases at the state level in Gujarat has to note this increase. In addition, the watershed programmes of the government would also have contributed their share to yields in the rainfed cotton plots too.

Seed Sources: Another factor that is often neglected in most analyses on cotton yields is the shift in seed source – from v arieties to hybrids and from F2 seed sources to F1. Table 6, based on data e stimates provided by the department of

Table 6: Estimates of Cotton Area

(in hectares under varieties and hybrids in Gujarat)

Source: Information obtained from the Department of Agriculture, Government of Gujarat, 2008.

Economic & Political Weekly

INSIGHT

ties to hybrids in the state.

Haryana

800

While cotton varieties occupy around

Rajasthan

on 43.7% of cotton area just a few years

600

500

ago (2003-04) as per the estimates of the

department. Further, there is a doubling

of absolute area under cotton hybrids

Yield (kg/ha)

400

d uring this period.

A report by the CICR in 2007 points out that one clear impact of Bt cotton on I ndian agriculture appears to be the r eplacement of large tracts of varietal areas with Bt hybrids since the (transgenic) technology is available in India only in the form of hybrids (Khadi 2007).

To take an example of another state, in Andhra Pradesh, cotton varieties were e stimated to be on 28% of cotton land in the state in 2002-03, the year of introduction of Bt cotton. Data obtained from the department of agriculture shows that the land under cotton varieties has come down to 1.3% by 2007-08.

Table 7: Cotton Area under Varieties, Non-Bt Cotton Hybrids and Bt Cotton Hybrids, Andhra Pradesh (2001-02 to 2007-08, area in lakh hectares)

Varieties (Non-Bt) Bt Cotton Total Varieties Hybrids Percentage

2001-02 3.32 7.76 11.08 30.0

2002-03 2.25 5.74 0.04 8.03 28.0

2003-04 2.34 5.98 0.05 8.37 28.0

2004-05 1.63 9.43 0.73 11.79 13.8

2005-06 0.16 6.93 3.24 10.33 1.5

2006-07 0.1 3.05 6.57 9.72 1.0

2007-08 0.15 1.17 10.01 11.33 1.3

Source: RTI data from Department of Agriculture, Government of Andhra Pradesh.

Apart from the picture related to “irrigated vs unirrigated” cotton and “varieties vs hybrids” in terms of seed source, the r eport from the CICR director also states that

Bt cotton seems to have reduced the overall quantity of insecticide substantially only in some parts of the country coupled with spectacular yield increases reported from Gujarat, while the rest of the states have been showing mixed results despite increase in area u nder Bt cotton.

The figure gives a picture of the cotton yields in different states of India. Andhra Pradesh, Gujarat, Karnataka, Madhya Pradesh, Maharashtra and Tamil Nadu have been cultivating Bt cotton from 2002 officially. The figure illustrates what Khadi says about the spectacular yield

i ncreases in Gujarat and mixed results in other states.

Interestingly, in the very first year of Bt cotton’s commercial cultivation in India (2002-03), the largest survey on performance was done in the state of Andhra Pradesh, by the department of agriculture. While there were 6,949 farmers who went in for Bt cotton in 2002-03 in Andhra Pradesh as per the sales figures of M onsanto-Mahyco, the department’s s urvey covered 3,709 of these farmers. This continues to be the largest percentage sample of Bt cotton performance in the country to this day. A whopping m ajority of the respondents in this survey (71%) reported low yields with Bt cotton.

Another factor worth analysing with r egard to current cotton cultivation in India is the use of chemical fertilisers. Increased use of fertilisers on Bt cotton is observed from various informal and media reports in addition to agriculture university recommendations as in the case of the recommendation by the Acharya N G Ranga A griculture University, Hyderabad (2007). There are no indications of yield analysis of cotton factoring in this increased use of chemical fertilisers on cotton.

Moving to another state, Madhya Pradesh posted impressive yields of 612.7 kg/ha per hectare on an average, for six years between 1996 and 2002 (six years prior to introduction of Bt cotton in the country), as per the Cotton Corporation of India’s records and such impressive results were without the help of GM technology. In 1997-98, Madhya Pradesh’s yields were a record 740 kg/ha – there was of course no Bt cotton then. However, in the six years after the introduction of Bt cotton, cotton yields in Madhya Pradesh have gone down to 518.3 kg/ha on a six-year-average.

In Conclusion

Most farmers as well as good agricultural scientists understand that crop yields are a complex phenomenon and depend on a variety of factors. However, the hype around Bt cotton and the cause-effect r elationship being attributed to Bt cotton and cotton yields in India required an article of this sort. This article was written only to explore the myth of yields behind GM crops like Bt cotton and not because the author believes that yields can be a simplistic linear function of one technology.

A matter of great concern is the lack of analysis of the current agrarian crisis in India (and using that understanding to predict the implications from risky technologies like GE), where farmers who have been relentlessly pushed towards treadmill technologies find that their costs of cultivation are ever-increasing, the productivity of their resources is constantly being eroded and despite increased yields, are left with dismal incomes of Rs 2,115 on an average! Going by various parameters to study the agrarian crisis, it is clear that yield increases from intensive agriculture models have not translated into livelihood improvements for farmers. This is something that the Kisan Policy of the National Commission of Farmers (NCT) (“Serving Farmers, Saving Farming”) headed by M S Swaminathan notes expressly. The NCF document states that the Kisan Policy attempts to “get the focus of our agricultural